The World Bank's indicator series gives Eritrea a population of 3,607,003 for 2025. Seven significant figures. The United Nations Statistics Division's register of census dates carries a dash against Eritrea for the 1990 round, the 2000 round, the 2010 round and the 2020 round, and the key at the foot of that table defines a dash as 'No census taken or planned in the period indicated'. Both statements are published by the international statistical system, and both are correct.
Silence Index — D-180
Quiet, but findable
Ranks 64 of 99 scored dossiers — quieter than 36% of the others we have scored. A low score is not a warning; it describes who else will be standing next to you.
How the total is weighted. These are the fixed weights of the four inputs, the same on every dossier — not this place’s scores. We do not publish per-input bars, because the per-input values were never recorded article by article and inventing four plausible numbers that sum to the total would be a fabrication dressed as a chart.
This dossier was scored before 2 September 2026, when the fourth input became night darkness. It keeps the input it was reasoned with until it is re-scored at its next source re-check; the methodology page records the change.
The index is a desk instrument scored by this publication, not a field measurement. How it is calculated, and where it fails.
What this dossier is built from
Nobody from this publication has been to Eritrea, and this is desk research from beginning to end. What follows is assembled from the UN Statistics Division's register of census dates and the metadata form Eritrea's own National Statistics Office filed with the UN Demographic Yearbook; the World Bank's indicator API, queried directly; the methodology report behind World Population Prospects 2024; two household survey reports published in Asmara, in 2003 and 2013; the Ethiopian census office's account of its 1984 enumeration; UNHCR's population statistics; the WHO and UNICEF immunisation estimates; UNESCO's record for Asmara; UNDP's country programme documents and the evaluation of them; and the Eritrean Ministry of Information's own country profile.
This is not an article about a country. It is an article about a number, and about the four empty cells in a table that the number stands in front of. It is not a mystery. A mystery is a thing whose shape nobody knows. This gap has a shape, a date, a method and two named holders: an office in Asmara that has never run a census, and a table in New York that records, round by round, that none was run.
The register, and what a dash means
The UN Statistics Division maintains a table titled World Population and Housing Census Programme – Census dates for all countries. It is not an analysis. It is a ledger. Each country occupies a row, each decadal round a column, and in each cell sits either a date or a symbol. The version we read was last updated on 3 February 2026.
The neighbouring rows are unremarkable. Ethiopia carries 11 October 1994 and then 29 May to 7 June 2007. Djibouti carries 29 May to 12 June 2009, then 20 May 2024. Sudan carries a population-only census on 15 April 1993 and a full one from 21 April to 6 May 2008.
Eritrea's row carries four dashes. 1990 round: dash. 2000 round: dash. 2010 round: dash. 2020 round: dash. The key beneath the table gives the dash exactly one meaning — "No census taken or planned in the period indicated". In the 2030 round column, covering 2025 to 2034, there is a bracketed ellipsis, which the same key defines as "It is expected that a census will be held during the decade". Expected. Not scheduled, not funded, not dated.
Eritrea is not alone in the shape of that row. Afghanistan, Lebanon and Western Sahara carry dashes across the same four rounds — but Afghanistan's last census was 1979 and Lebanon's was 1932, and both of those are dates. Eritrea, a UN member state since 28 May 1993, has no date of its own.
What the number is made of
The honest way to interrogate a statistic is to ask the body that produced it what it did, and the international system provides a form for exactly that: each year, statistical offices file a metadata questionnaire with the UN Demographic Yearbook explaining how their estimates were built. Eritrea's National Statistics Office filed one. It is short, and candid.
The base is the 2002 population quick count. The type of estimate is recorded as "Post-censal, after the quick of population in Eritrea done in 2002", and the source reads "Estimated based on the 2002 population quick count and the population dynamics results of the 1995, 2002, and 2010 DHS surveys". The estimates were produced on 15 April 2022 and run to 2030, de facto, status final revised.
Section 7 asks how the base was carried forward. It offers a menu. Option two reads "Applying statistics of births and deaths (natural increase), and migration". Option three reads "Applying statistics of births and deaths (natural increase), but no account taken of migration". The only option carrying a completed free-text specification is option three, and that specification reads: "Assumption on future course of fertility and mortality was made based on the 1995, 2002, and 2010 EDHS findings".
That is the whole machine. A quick count taken in 2002, aged forward twenty-three years on fertility and mortality curves fitted to household surveys, the most recent fielded in 2010, with no migration term. UNHCR's database records 551,296 refugees, 111,762 asylum-seekers and 6,354 others in refugee-like situations originating from Eritrea at the end of 2024 — 669,412 people, roughly one in six of the published national total. Whether they sit inside the 3,607,003 or outside it is not a question the method can answer, because the method does not contain the term.
What popular coverage gets wrong
The version in general circulation is a flat sentence: Eritrea has a population of about 3.6 million. It appears without qualification in country profiles, in the front matter of guidebooks, in travel writing about the Red Sea coast and in the standing furniture of encyclopaedias. UNFPA's world population dashboard gives 3,600,000 for 2025. UN Data gives 3,607 thousand. The World Bank gives 3,607,003.
None of them is wrong about what it publishes. They are wrong in the reading they invite: that somebody counted.
UN Data footnotes its own figure. The letter beside 3,607 resolves to "Projected est. (medium fertility variant)". So does the letter beside the fertility rate of 3.6, the share under fifteen at 37.6 per cent, the share over sixty at 6.5 per cent, life expectancy at 71.2 years for women and 67.0 for men, and under-five mortality at 32.6 per thousand. The entire demographic portrait of the country, in the UN's own profile, carries one footnote.
The World Bank's definition says the values "are midyear estimates" on a de facto basis, sourced to World Population Prospects and to national statistical offices. The methodology report behind World Population Prospects 2024 is more specific. It lists twenty-three countries whose most recent census data predate 2009, with dates in brackets, and the list runs "Lebanon (1932), Afghanistan (1979), Democratic Republic of the Congo (1984), Eritrea (1984), Somalia (1987)". Of the 2,025 censuses conducted worldwide between 1950 and 2023, the Population Division used 1,910. None was Eritrean.
The 1984 problem, and what the office says
That bracketed 1984 requires care, and it is why this dossier does not use the phrase "never counted" without qualification.
Ethiopia held a Population and Housing Census on 9 May 1984, when Eritrea was administered as Ethiopian territory. The enumeration was real and large: 34,500,972 people on a de jure basis. It was also incomplete, and the census office said so. Its source documentation states that "The census has not covered low land areas with nomadic population; rural areas of Asseb, Tigray and Eritrea and some of the urban centers in Tigray, Eritrea and other regions." The unenumerated population was put at 8,115,904, so the count reached roughly 81 per cent of the country.
Eritrean territory was therefore partially enumerated once, in 1984, by another state's census office, with its rural areas and some of its towns explicitly excluded. That is the entire enumerative record, and it explains how World Population Prospects can write "Eritrea (1984)" while the census register shows four dashes: the register begins its accounting with the 1990 round, which opens in 1985. The defensible formulation is the one UNDP's own independent evaluation uses — that Eritrea has not conducted a population census since independence.
The most direct source on this is not international. It is Eritrean, and it is in print twice. The Eritrea Demographic and Health Survey 2002, published in Asmara in May 2003 by the National Statistics and Evaluation Office with ORC Macro, opens its population section without equivocation: "No population census has ever been carried out in Eritrea. As a result, there are no reliable estimates of the population currently residing in Eritrea or the population of Eritreans living abroad, many of whom are potential returnees." It adds that, based on a population count, the Ministry of Local Government estimated the total at about 3.2 million as of 2001.
Eight years later the Eritrea Population and Health Survey 2010 — published in Asmara in August 2013 by the National Statistics Office with the Fafo Institute for Applied International Studies — repeats the finding almost word for word, and gives about 3.2 million again, this time as of 2010, citing the Ministry of National Development. Two official reports, nine years apart, the same round figure for two different years. We do not attempt to reconcile that. We note only that the World Bank's series gives 2,392,978 for 2002 and 2,945,186 for 2010, and that the Eritrean Ministry of Information's public country profile still reads "Population size: 3.56 million (2002)" — which, set against the World Bank's value for the same year, is a difference of more than 1.1 million people.
What is not known, and at what resolution
It is worth being exact about the shape of the gap.
- How many people. No enumerated national total exists for Eritrean territory at any date since independence in 1993. Every figure in circulation is a projection from the 2002 quick count, or an estimate attributed to a Ministry of Local Government count whose report we could not obtain.
- In which regions. Eritrea has six zobas — Anseba, Debub, Debubawi Keih Bahri, Gash-Barka, Maekel and Semenawi Keih Bahri — named as such in the National Statistics Office's own 2010 report. We found no enumerated population for any of them. Subnational figures exist only as model outputs distributed as humanitarian planning layers.
- Of what ages. The age structure is a projection. UN Data's 0–14 and 60+ shares for 2025 both carry the medium-variant footnote. No published age-sex distribution for Eritrea rests on an enumeration.
- How many live in Asmara. UN Data gives 928,800, footnoted to 2019. The GeoNames gazetteer gives 563,930. Neither cites a count.
- How poor. The World Bank states that "Poverty appears to have remained widespread in Eritrea, but the lack of data limits available quantitative evidence", its most recent survey evidence being a 70 per cent headcount from 1996/97. Its poverty series holds null in all sixty-six annual cells from 1960 to 2025.
- How well the statistical system works. The World Bank's Statistical Performance Indicators score is null for Eritrea in every year, where 2023 gives Ethiopia 62.7, Somalia 45.7 and South Sudan 23.8. The UN's directory of national statistical office websites runs, in its Africa section, from Djibouti straight to Egypt.
What the gap costs
A denominator is not an abstraction. It is the bottom of every rate a country is governed by, and when it is uncertain the uncertainty propagates into decisions about real people.
Take immunisation, measured as doses over target population. The WHO and UNICEF estimates for Eritrea put DTP3 coverage at 95 per cent for 2023 and BCG at 97 per cent, matching what the country officially reports. The administratively reported figures — doses divided by the national target population — run at 78 and 71 per cent for the same year, and have sat around twenty points below the official series for a decade. The grade-of-confidence note gives part of the mechanism: coverage is "recalculated with independent denominator from World Population Prospects". Two denominators, one national and one modelled, produce two answers about the same children. We cannot say which is closer to the truth, and neither, on this record, can anyone else. The same profile records "No nationally representative independent assessment for the most recent 5 annual birth cohorts".
Take refugee status determination and resettlement planning, where the denominator decides what a flow means. UNHCR's operational portal for Ethiopia recorded 171,964 Eritrean refugees and asylum-seekers on 31 July 2026 — 15.3 per cent of the 1,127,629 people that country hosts. Whether the total outflow represents four per cent of Eritrea's population or seven is a question of humanitarian consequence, and the arithmetic to settle it does not exist.
Take aid allocation, which is per capita almost everywhere. UNDP's draft country programme for 2027–2031 states that "Eritrea's population is approximately 3.7 million, growing at a rate of 2.3 per cent", with no source attached — a fourth figure, differing from the World Bank's by some 93,000 people — and flags the "absence of readily available national data and a constrained national monitoring and evaluation ecosystem" as a risk to its own delivery. UNDP's independent evaluation of the previous programme was blunter: national accounts and GDP figures "relied on unofficial estimates", inflation data "were limited to Asmara", and no full balance of payments accounts existed. Eritrea does not appear among the countries analysed by the Integrated Food Security Phase Classification, which covers Djibouti, Ethiopia, Kenya, Somalia, Uganda and Tanzania.
The one list, and what is moving now
There is a single exception in the record, and it is not a census. In April 1992 the Provisional Government of Eritrea established a Referendum Commission and, by the state's own later account, "a computerised registration of voters was also organized" across Eritrea, Ethiopia, Sudan and more than forty other countries. The Ministry of Information puts the result at over 1.1 million registered eligible voters, a turnout of 98.5 per cent across the poll of 23 to 25 April 1993, and 99.81 per cent in favour of independence. A voter register is not a population: it excludes children and it deliberately included a diaspora. But it remains the only name-by-name list of Eritreans the state has ever assembled, and it was assembled to answer a political question rather than a demographic one.
Something is moving now, and it should be reported as carefully as the absence. In 2018 UNFPA's deputy executive director for programmes, Dereje Wordofa, said in Asmara that "UNFPA is ready to support the country to undertake a Census and Demographic and Health Survey". Nothing followed on the register. In July 2020 the European Commission and UNDP signed a four-year project, SENSS, worth €4.75 million and €1 million, to strengthen the national statistics system with the Ministry of National Development and the National Statistics Office.
In May 2026 UNDP reported that the National Statistics Office had completed the fourth round of the Eritrea Population and Health Survey, EPHS 2025, with technical support from Mikro International Consultancy Service and the UN Economic Commission for Africa. It reached 9,794 households across 405 enumeration areas in all six regions, with a 98.4 per cent household completion rate. Nearly 400 tables across twenty chapters had been produced, with printing and dissemination scheduled for 2026. We have not been able to obtain the report. It is a survey, not a census, and it will not produce a national total — but it is the first of its kind in fifteen years, and the register's 2030 column now holds an ellipsis rather than a dash, which in the grammar of that table is the difference between "no census taken or planned" and "expected during the decade".
Why the Silence Index is 67
Asmara is a capital with an international airport, so physical remoteness contributes almost nothing. Access friction does the heavy lifting: all foreign nationals — including embassy staff — must hold a permit to travel more than twenty-five kilometres beyond Asmara, applied for at one of two Ministry of Tourism offices, taking several days and, in the FCDO's phrasing, "not always successful". Credit cards are accepted nowhere in the country and there is no mobile data or roaming for visitors. Footfall is genuinely low: the World Bank's arrivals series peaked at 417,000 in 1996, stood at 142,000 in 2016 and has been null every year since. The fourth input is unresolved questions, and it is the one the dossier turns on — there is no working town on earth whose basic demography is less settled. What holds the score below the seventies is that this is a lived-in city of half a million people or more. Administratively hard to reach and statistically unmapped is a different thing from empty.
The ledger and the city
Asmara was inscribed by UNESCO in 2017, under criteria (ii) and (iv), as "Asmara: A Modernist African City" — 481 hectares of property inside a 1,203-hectare buffer zone, described in the Outstanding Universal Value statement as "an exceptionally well-preserved example of a colonial planned city, which resulted from the subsequent phases of planning between 1893 and 1941, under the Italian colonial occupation". The original urban layout survives in its entirety, protected by a Historic Perimeter established in 2001 and a municipal moratorium on new construction within it. The cinemas — Impero, Roma, Odeon, Capitol, Hamasien — are named in the citation.
It is a strange pairing. The buildings of Asmara have been surveyed, mapped, buffered and inscribed to the hectare by an international body. The people living among them have not been counted. Out on the Dahlak archipelago, the World Database on Protected Areas holds a record for "Dahlak Island PA" whose reported area reads, in the field where a number should be, "Not Reported".
There is nothing occult in any of this. Censuses are expensive and logistically enormous, and a country that fought for independence until 1993 and a border war until 2000 had reasons to defer one that any statistician would recognise. What the record shows is the deferral itself, repeated across four decadal rounds, and the quiet substitution of a model for a count. The number will keep being published to the unit, and repeated flatly. Behind it, four cells will keep holding a symbol that means exactly what the key says it means.
Corner Codex — D-180
- Country
- State of Eritrea; UN member since 28 May 1993
- Capital
- Asmara, 15.3381°N 38.9317°E (GeoNames)
- Published population
- 3,607,003 for 2025 (World Bank, indicator SP.POP.TOTL)
- What that number is
- A post-censal projection from a 2002 quick count, no migration term
- Censuses on the UN register
- None. A dash in the 1990, 2000, 2010 and 2020 rounds
- Meaning of the dash
- 'No census taken or planned in the period indicated'
- Next expected
- 2030 round (2025–2034), marked only as expected during the decade
- The state's own words
- 'No population census has ever been carried out in Eritrea' (NSO)
- Government's published figure
- 3.56 million, dated 2002, on the Ministry of Information site
- Eritreans abroad
- 669,412 refugees, asylum-seekers and others, end-2024 (UNHCR)
- Statistical capacity score
- Null for every year 1996–2025 (World Bank SPI)
- Last poverty survey
- 1996/97, 70% headcount (World Bank)
- Silence Index
- 67 / 100
The Traveler's Panel · checked
Changing information. Prices, permits, seasons and road access change, sometimes at short notice. Everything here is a starting point for your own confirmation — not a quote, and not a guarantee. Verify with the operator and the relevant official body before you book.
| The published figure | 3,607,003 for 2025 (World Bank). Seven significant figures |
| What it rests on | A 2002 quick count, aged forward on 1995/2002/2010 survey curves |
| Migration term | None. The filed method line reads 'no account taken of migration' |
| UN census register | Dash in the 1990, 2000, 2010 and 2020 rounds; expected in the 2030s |
| The office's own words | 'No population census has ever been carried out in Eritrea' |
| Travel beyond Asmara | Permit required past 25 km, from the Ministry of Tourism |
| Advisory | US Level 2, 31 July 2023. Credit cards accepted nowhere; no data roaming |
| Last arrivals figure | 142,000 in 2016. Null every year since |
Getting there — Asmara International Airport is the only realistic arrival point; everything beyond 25 km needs a permit
- Everyone arrives at Asmara. There is a visa requirement for all nationalities and the FCDO directs applicants to the Eritrean Embassy rather than to any online process; build the timeline around the visa rather than around flights, and expect the answer to come from Asmara.
- The permit is the real gate, and it is not a formality. The US State Department states that all foreign nationals, including its own embassy officials, must obtain permits for travel more than twenty-five kilometres outside Asmara. Applications go to one of two Ministry of Tourism offices, on Harnet Avenue and Airport Road. The FCDO's framing is that these take several days and are not always successful.
- That single rule shapes any itinerary. Massawa, Keren, the Dahlak boats and the Qohaito escarpment all sit outside the twenty-five kilometre radius, which means each of them is a separate piece of paper rather than a day's driving decision.
- The islands carry their own layer. The FCDO records that vessels have been impounded and crew detained for entering Eritrean waters or landing on Eritrean islands without permission, and advises mariners to obtain permissions and entry visas in advance. Arriving at the Dahlaks by your own keel is not a plan.
- Carry cash and expect to account for it. Foreign currency of USD 10,000 or more must be declared on entry and reconciled on exit through state exchange or official hotel spending, and no more than 1,000 nakfa may be taken out. Credit cards are accepted nowhere in the country and exchange goes through Himbol branches.
- Assume you are offline. The State Department notes that it is very difficult for a tourist to obtain a SIM card, and that there is no data service or roaming. Photographing government buildings and military installations is illegal, and even the tank graveyard on the edge of Asmara requires a Ministry of Tourism permit.
Indicative costs — verify before booking
| Entry | Visa required for all nationalities; fees are set by Eritrean missions and we found no published tariff we could verify |
| Travel permit | No published fee we could confirm. The cost is time — several days per application, with no guarantee of approval |
| What we will not quote | Tour prices. Every figure we found came from operators selling the itinerary, and none from a state schedule |
| Money on the ground | Cash only. Credit cards are accepted nowhere in Eritrea; exchange at state Himbol branches |
| Currency rules | Declare USD 10,000 or more on entry and reconcile it on exit; maximum 1,000 nakfa out |
| The honest line | This is not a place to price. It is a place to obtain permission for, and the permission is the expensive part |
What to pack
- Cash, in a hard currency, in an amount you are prepared to declare and account for on the way out. There is no card network to fall back on anywhere in the country.
- Printed copies of every permission you hold — visa, travel permit, photography permit. This is a paperwork state and the paperwork is checked at road controls outside Zoba Maekel.
- Offline maps and offline everything. No tourist data service, no roaming, and a SIM card that is very difficult to obtain.
- Clothing for two climates in one day. Asmara sits above 2,000 metres on the highland plateau; Massawa is at sea level on the Red Sea, roughly a hundred kilometres and two thousand vertical metres away.
- A camera you are willing to put away. Government buildings and military installations are off-limits by law, and the consequences described by the State Department run from a warning to detention.
Where to stay
Almost everyone stays in Asmara, because almost everyone must: the twenty-five kilometre permit radius makes the capital the default base and every other night a separate application. The city sits above 2,000 metres on the highland plateau, and its centre is the 481-hectare property UNESCO inscribed in 2017 inside a 1,203-hectare buffer zone, protected since 2001 by a Historic Perimeter and a municipal moratorium on new construction within it — which is the reason the hotel stock is largely mid-century and the reason it has not been replaced. Beyond Asmara, Massawa on the Red Sea and Keren in the Anseba highlands are the two towns with any meaningful accommodation; the gazetteer records Massawa at 23,100 people and Keren at 74,800, though those are gazetteer values resting on no enumeration, like every other population figure in this dossier. Assab, far south at 13.0°N, is effectively a separate country to reach. There is no visitor infrastructure on the Dahlaks that we could verify from any official source, and the State Department's assessment of the sector as a whole is that tourism infrastructure is minimal and tourists participate at their own risk.
Safety & responsible travel
- Do not repeat 3.6 million as a counted figure. It is a projection from a 2002 quick count with no migration term, and both the National Statistics Office and UNDP's own evaluation have said in print that no census has been carried out since independence.
- Be equally careful in the other direction. 'Nobody knows how many people live in Eritrea' is closer to the record than '3.6 million', but the country is not unmeasured — four household surveys have been fielded since 1995, the most recent in 2025, and their results are real even though they cannot produce a national total.
- Respect the permit system rather than testing it. It applies to diplomats as well as to visitors, and the people who bear the cost of a foreigner improvising are usually the local driver, guide or host, not the foreigner.
- Photograph buildings, not installations, and ask before photographing people. Asmara's modernist fabric is the reason the city is inscribed, and it is also where several thousand people live and work.
- On the Dahlaks, treat the reef as unsurveyed, because it effectively is. The World Database on Protected Areas holds a record for Dahlak Island PA with its reported area entered as 'Not Reported', and none of Eritrea's eleven protected areas has a management effectiveness evaluation.
- If you are writing about the country afterwards, cite the register rather than the summary. The UN census dates table takes about a minute to read and settles the question that most country profiles get wrong in their first sentence.
Nearby, and quieter
Massawa, roughly a hundred kilometres and two thousand vertical metres down from Asmara, is the port from which the Dahlak archipelago is reached — a group of islands that GeoNames records under twenty-five separate names, with Dahlak Kebir the largest and a recorded settlement population of 2,500, and which appears on neither UNESCO's World Heritage list nor Eritrea's tentative list. Keren, north-west in Anseba, is the second town of the highlands. South of Asmara, near Adi Keyh, lies Qohaito, the escarpment cultural landscape Eritrea placed on its UNESCO tentative list in 2011 and has not advanced since. Nak'fa, deep in the northern mountains and lending its name to the currency, has a gazetteer entry with coordinates and no population attached to it at all — which is, in miniature, the subject of this dossier. Across the border, Ethiopia's operational portal counted 171,964 Eritrean refugees and asylum-seekers on 31 July 2026, which is the other reason the denominator matters.
Getting there, in order
- 1Massawa
- 2Dahlak Archipelago
- 3Assab
- 4Asmara
Each leg is a separate booking and a separate chance to be weathered out. Build spare days into the chain, not onto the end of it.
Where it is
Eritrea — Asmara, the highlands and the Red Sea coast · 15.3381°N 38.9317°E · Open in Google Maps · OpenStreetMap
Key locations
- Asmara — capital; UNESCO ref 1550, 481 ha inside a 1,203 ha buffer. Gazetteer population 563,930
15.3381°N 38.9317°E - Massawa — Red Sea port and the gateway to the Dahlaks; gazetteer population 23,100
15.6081°N 39.4744°E - Keren — seat of Anseba zoba, north-west of Asmara; gazetteer population 74,800
15.7779°N 38.4511°E - Nak'fa — northern highland town; gazetteer entry carries coordinates but no population
16.6672°N 38.4764°E - Dahlak Kebir — largest island of the archipelago; the settlement is recorded at 2,500 people
15.6597°N 40.1142°E - Dahlak Archipelago — island group, Northern Red Sea; the WDPA record's reported area reads 'Not Reported'
15.8333°N 40.2000°E - Assab — southern Red Sea port, seat of Southern Red Sea zoba; gazetteer population 21,300
13.0092°N 42.7394°E
Coordinates are for orientation and are accurate to roughly the width of the feature named, not to a doorway. Several of these points are inside protected areas or need a permit, a guide or a boat — read the access notes above before you plan a route to any of them.
Sources & further reading — checked
- UN Statistics Division — World Population and Housing Census Programme, 'Census dates for all countries' (last update 3 February 2026). The register itself: Eritrea carries a dash in the 1990 (1985–1994), 2000 (1995–2004), 2010 (2005–2014) and 2020 (2015–2024) round columns and an ellipsis for the 2030 round; the key defines '–' as 'No census taken or planned in the period indicated' and '(...)' as 'It is expected that a census will be held during the decade'; Afghanistan, Lebanon and Western Sahara carry the same four dashes, while Ethiopia carries 11 October 1994 and 29 May–7 June 2007, Djibouti 29 May–12 June 2009 and 20 May 2024, and Sudan (P) 15 April 1993 and 21 April–6 May 2008
- UN Statistics Division — Demographic Yearbook, 'Questionnaire on Population Estimates, Part 2: Metadata', Eritrea, 2025 (filed by the National Statistics Office). Eritrea's own metadata: source 'Estimated based on the 2002 population quick count and the population dynamics results of the 1995, 2002, and 2010 DHS surveys'; type of estimate 'Post-censal, after the quick of population in Eritrea done in 2002'; produced 15/04/2022 covering 2002–2030; de facto present population; and the completed method line reading 'Applying statistics of births and deaths (natural increase), but no account taken of migration'
- UN Statistics Division — 'Partners', National Statistical Offices Websites directory. The UN's own directory of national statistical office websites: the Africa section runs from Djibouti (Ministère de l'Economie, des Finances et de la Planification) straight to Egypt, and the string 'Eritrea' does not appear anywhere on the page
- World Bank — Population, total, Eritrea (indicator SP.POP.TOTL), API query. The seven-significant-figure series: 3,607,003 for 2025, 3,535,603 for 2024, 3,470,390 for 2023, 3,409,447 for 2022, 3,350,183 for 2021 and 3,291,271 for 2020, with no null years across that span
- World Bank — Population, total, Eritrea, 1993–2010 (indicator SP.POP.TOTL), API query. The earlier values used for comparison against Eritrean official figures: 1,856,006 for 1993, 2,392,978 for 2002 and 2,945,186 for 2010 — the last two set against the Ministry of Information's 3.56 million for 2002 and the National Statistics Office's about 3.2 million for 2010
- World Bank — indicator metadata for SP.POP.TOTL (long definition and source organisations). The definition: 'Total population is based on the de facto definition of population, which counts all residents regardless of legal status or citizenship. The values shown are midyear estimates', sourced to World Population Prospects, national statistical offices, Eurostat and the Population and Vital Statistics Report
- World Bank — Statistical performance indicators (SPI): Overall score, Eritrea (IQ.SPI.OVRL). Thirty annual records for Eritrea from 1996 to 2025, every one of them null — the World Bank publishes no statistical capacity score for the country at all
- World Bank — SPI overall score, 2023, comparison set (Ethiopia, Somalia, South Sudan, Afghanistan, Eritrea). Ethiopia 62.67, Afghanistan 56.75, Somalia 45.68 and South Sudan 23.84 for 2023, against null for Eritrea — establishing that the absence of a score is specific rather than a general feature of low-income or conflict-affected states
- World Bank — Poverty headcount ratio series for Eritrea (SI.POV.DDAY query). Sixty-six annual records for Eritrea from 1960 to 2025 under 'Poverty headcount ratio at $3.00 a day (2021 PPP)', with a null value in every single cell
- World Bank — Eritrea country overview. The statements that 'Poverty appears to have remained widespread in Eritrea, but the lack of data limits available quantitative evidence' and that 'The most recent available survey data from 1996/97 indicate a 70% poverty rate', alongside a maternal mortality ratio of 485 per 100,000 live births and 52% of under-fives underweight
- World Bank — International tourism, number of arrivals, Eritrea (ST.INT.ARVL), API query. Arrivals peaking at 417,000 in 1996, falling to 57,000 in 1999, recovering to 142,000 in 2016, and null for every year from 2017 to 2025 — the basis for this dossier's refusal to estimate current visitor numbers
- UN Population Division — World Population Prospects 2024: Methodology of the United Nations population estimates and projections. The list of twenty-three countries whose most recent available census data predate 2009, running 'Lebanon (1932), Afghanistan (1979), Democratic Republic of the Congo (1984), Eritrea (1984), Somalia (1987)'; Eritrea grouped with South Sudan and Micronesia as having fertility data no later than 2010; and 2,025 censuses conducted worldwide 1950–2023 of which 1,910 were used
- UN Data — Country profile, Eritrea. Population 3,607 thousand for 2025 footnoted 'Projected est. (medium fertility variant)', with the same footnote on fertility (3.6), the 0–14 share (37.6%), the 60+ share (6.5%), life expectancy (71.2/67.0) and under-five mortality (32.6); capital city population 928.8 thousand footnoted to 2019; surface area 121,144 km²; UN membership 28 May 1993
- UNFPA — World Population Dashboard, Eritrea. A total population of 3,600,000 for 2025, sourced to 'World Population Prospects 2024 revision. United Nations Population Division, 2024' — the rounded form of the same modelled figure
- National Statistics and Evaluation Office [Eritrea] and ORC Macro — Eritrea Demographic and Health Survey 2002, final report FR137 (Asmara, May 2003). Section 1.2: 'No population census has ever been carried out in Eritrea. As a result, there are no reliable estimates of the population currently residing in Eritrea or the population of Eritreans living abroad, many of whom are potential returnees. However, based on a population count, the Ministry of Local Government estimated the total population of Eritrea to be about 3.2 million as of 2001.' Sample of 9,389 households and 8,754 women aged 15–49
- National Statistics Office [Eritrea] and Fafo Institute for Applied International Studies — Eritrea Population and Health Survey 2010, final report (Asmara, August 2013). Section 1.1.4: 'No population census has ever been carried out in Eritrea. However, based on a population count by the Ministry of Local Government and NSO estimates, the total resident population of Eritrea was about 3.2 million as of 2010 (MND, 2010).' Sample of 34,423 households across 900 clusters, fieldwork January to early July 2010, and the six zobas named as Anseba, Debub, Debubawi Keih Bahri, Gash-Barka, Maekel and Semenawi Keih Bahri
- IPUMS International / Transitional Government of Ethiopia, Office of the Population and Housing Census — source documentation for the 1984 Population and Housing Census of Ethiopia. Census day 9 May 1984; 34,500,972 people enumerated on a de jure basis, about 81% of the country, with 8,115,904 unenumerated; and the coverage statement that 'The census has not covered low land areas with nomadic population; rural areas of Asseb, Tigray and Eritrea and some of the urban centers in Tigray, Eritrea and other regions'
- World Bank Microdata Library — Ethiopia, 1984 Population and Housing Census, IPUMS subset. The catalogue record confirming a de jure census with census day 9 May 1984 at awraja level, and the explicit warning that the 'sample is not representative of the country-level population. Due to famine and conflict, parts of the country are missing from the sample'
- UNHCR — Refugee Data Finder API, population by country of origin Eritrea, 2024. End-2024 totals for people originating from Eritrea: 551,296 refugees, 111,762 asylum-seekers and 6,354 others out of country — 669,412 in all, the figure this dossier sets against a national estimate built without a migration term
- UNHCR — Operational Data Portal, Ethiopia country page. As at 31 July 2026, 171,964 Eritrean refugees and asylum-seekers in Ethiopia, 15.3% of the 1,127,629 refugees and asylum-seekers the country hosts, the third-largest group after South Sudanese (42.9%) and Somalis (32.2%)
- WHO and UNICEF — Estimates of national immunization coverage, Eritrea, 2023 revision. DTP3 at 95% estimated and official against 78% administrative for 2023, BCG at 97% against 71% administrative, a twenty-point gap sustained for a decade; the grade-of-confidence definition citing coverage 'recalculated with independent denominator from World Population Prospects: 2022 revision'; and the note 'No nationally representative independent assessment for the most recent 5 annual birth cohorts'
- UNDP — Independent Country Programme Evaluation, Eritrea. The findings that 'Eritrea has not conducted a population census since independence' and that 'the last demographic and health survey was carried out in 2010, resulting in outdated demographic data'; that 'National accounts data and GDP figures relied on unofficial estimates. Inflation data were limited to Asmara, and no full balance of payments accounts existed'; and that many CPD indicators lacked baselines or targets
- UNDP Executive Board — draft country programme document for Eritrea, 2027–2031 (DP/DCP/ERI/5). The unsourced statement that 'Eritrea's population is approximately 3.7 million, growing at a rate of 2.3 per cent'; the risk flagged as the 'absence of readily available national data and a constrained national monitoring and evaluation ecosystem'; investment in the national data ecosystem 'after a prolonged gap'; and no mention of a census, past or planned
- UNDP Eritrea — 'Driving Data with Digital Technology: Eritrea Strengthens Data Systems with Landmark Survey' (8 May 2026). The Fourth Round Eritrea Population and Health Survey (EPHS 2025) run by the National Statistics Office with Mikro International Consultancy Service and UNECA; 9,794 households across 405 enumeration areas in all six regions; a 98.4% household completion rate; the move to computer-assisted personal interviewing; and nearly 400 tables across 20 chapters with printing and dissemination scheduled for 2026
- European External Action Service, EU Delegation to Eritrea — Strengthening Eritrea's National Statistics and Macroeconomic Statistics Systems (SENSS). A four-year project signed on 29 July 2020, co-funded by the European Commission (€4.75 million) and UNDP (€1 million), implemented with the Eritrean Ministry of National Development and the National Statistics Office to strengthen the national statistics system
- UNFPA East and Southern Africa Regional Office — 'UNFPA strengthening partnerships in Eritrea to sustain development gains' (20 October 2018). Deputy Executive Director for Programmes Dereje Wordofa stating in Asmara that 'UNFPA is ready to support the country to undertake a Census and Demographic and Health Survey' — an offer made eight years before the UN census register recorded any change
- Ministry of Information of the State of Eritrea — 'Eritrea at a Glance'. The government's own published country facts, including 'Population size: 3.56 million (2002)' with no source and no more recent figure, an area of 124,300 sq. km against UN Data's 121,144 km², and a stated 60% rural to 40% urban split
- Ministry of Information of the State of Eritrea — 'A Look Back on Eritrea's Historic 1993 Referendum' (25 April 2018). The state's account of the only national listing exercise it has run: a Referendum Commission established in April 1992, 'a computerised registration of voters', over 1.1 million registered eligible voters across Eritrea, Ethiopia, Sudan and more than forty other countries, a 98.5% turnout on 23–25 April 1993 and 99.81% in favour of independence
- UNESCO World Heritage Centre — Asmara: A Modernist African City (ref. 1550). Inscribed in 2017 under criteria (ii) and (iv); 481 ha of property with a 1,203 ha buffer zone at N15 20 7 E38 56 9; the Outstanding Universal Value statement calling it 'an exceptionally well-preserved example of a colonial planned city, which resulted from the subsequent phases of planning between 1893 and 1941, under the Italian colonial occupation'; the Historic Perimeter and construction moratorium since 2001; and the Impero, Roma, Odeon, Capitol and Hamasien cinemas
- UNESCO World Heritage Centre — States Parties, Eritrea. Eritrea's acceptance of the World Heritage Convention on 24 October 2001, one inscribed property (Asmara, 2017), and a single tentative-list entry, Qoahito Cultural Landscape, submitted in 2011 — the Dahlak islands appear on neither list
- UNEP-WCMC and IUCN, Protected Planet — Dahlak Island PA, Eritrea (WDPA 555721463). The World Database on Protected Areas record for a marine and coastal protected area in the Dahlak archipelago whose reported area is entered as 'Not Reported'
- UNEP-WCMC and IUCN, Protected Planet — country profile, Eritrea. Eleven protected areas recorded for Eritrea, none with a management effectiveness evaluation and no other effective area-based conservation measures
- Integrated Food Security Phase Classification — country analyses. The list of countries under IPC analysis, which includes Djibouti, Ethiopia, Kenya, Somalia, Uganda and the United Republic of Tanzania in eastern Africa and does not include Eritrea
- UK Foreign, Commonwealth and Development Office — Eritrea travel advice, safety and security. 'Foreign nationals must apply in advance for a permit to travel outside Asmara's surrounding province, Zoba Maekel', applied for by tourists at the Ministry of Tourism on Harnet Avenue, taking several days and 'not always successful'; the illegality of photographing government buildings and military installations; and the warning that vessels have been impounded and crew detained for entering Eritrean waters or landing on Eritrean islands without permission
- UK Foreign, Commonwealth and Development Office — Eritrea travel advice, entry requirements. A visa required to enter; foreign cash or travellers cheques of USD 10,000 or more to be declared and accounted for on departure or risk prosecution and a severe fine; a limit of 1,000 nakfa taken out of the country; and an exit permit required for dual nationals who enter on an Eritrean identity card
- US Department of State — Eritrea International Travel Information. 'All foreign nationals, including U.S. Embassy officials, are required to obtain permits for travel more than 25 km outside of Asmara', applied for at two Ministry of Tourism offices on Harnet Avenue and Airport Road; credit cards accepted nowhere in the country; no data service or roaming available to visitors; and 'Tourism industry infrastructure is minimal'
- US Department of State — Eritrea Travel Advisory. Level 2, Exercise Increased Caution, dated 31 July 2023, citing travel restrictions, limited consular assistance, landmines and wrongful detentions, and noting that US government employees must obtain special authorisation to travel outside Asmara
- GeoNames gazetteer — Asmara, Eritrea. Asmara as capital of a political entity at N15°20'17" E38°55'54" (15.3381°N, 38.9317°E) in Maekel, with a gazetteer population of 563,930 — set against UN Data's capital city figure of 928,800 footnoted to 2019
- GeoNames gazetteer — Massawa, Keren, Assab and Nak'fa, Eritrea. Massawa, seat of a first-order administrative division, at N15°36'29" E39°28'28" with a population of 23,100; used with the sibling queries for Keren (15.77792°N, 38.451074°E, 74,800), Assab (13.00917°N, 42.73944°E, 21,300) and Nak'fa (N16°40'02" E38°28'35", no population recorded)
- GeoNames gazetteer — Dahlak archipelago, Eritrea. Twenty-five records for the name, including the Dahlak Archipelago as an island group at N15°50' E40°12' in Northern Red Sea, Dahlak Kebir island at N15°39'35" E40°06'51", and Dehalak' Kebir as a populated place with a recorded population of 2,500
Official and scientific bodies are the authority for safety, regulatory and scientific claims. Commercial operators are used only for practical detail such as tour length, meeting points and indicative pricing — never for a safety or scientific statement.
How this was made
Desk research. Compiled from official bodies, scientific and conservation sources, established journalism and licensed operators. Not based on a personal visit.
Editor’s method note — what we checked, corrected and could not verify
- Desk research only. Nobody from this publication has been to Eritrea, and no part of this dossier rests on observation.
- Every URL cited was opened and read before citation. Sources that returned an error, a robots block or no machine-readable text are listed near the end of this note and are cited nowhere in the piece.
- We do not state a population for Eritrea in our own voice at any point. We report what each body publishes and attribute it: 3,607,003 for 2025 from the World Bank's indicator series; 3,600,000 from UNFPA's dashboard; 3,607 thousand from UN Data with a medium-variant footnote; approximately 3.7 million from UNDP's draft country programme, unsourced; 3.56 million dated 2002 from the Eritrean Ministry of Information; and about 3.2 million from the National Statistics Office, given for 2001 in one report and for 2010 in another. We print all of them and name all of them. We do not average them and we do not choose between them.
- On the migration term: the UN Demographic Yearbook metadata form is a scanned questionnaire and its checkbox states are not machine-readable to us. Our reading is that option 7.1.1(3) — 'Applying statistics of births and deaths (natural increase), but no account taken of migration' — is the selected one, because it is the only method line carrying a completed free-text specification. We state it that way in the body rather than as a tick we have seen.
- The phrase 'never counted' is avoided deliberately. Eritrean territory was partially enumerated in Ethiopia's census of 9 May 1984, and the Ethiopian census office recorded that the enumeration excluded rural Eritrea and some of its urban centres. That is why World Population Prospects can list 'Eritrea (1984)' while the UN census register shows four dashes from the 1990 round onward. The defensible formulation, and the one UNDP's own evaluation uses, is that Eritrea has not conducted a population census since independence.
- We made a specific effort to open Eritrean government statistical sources rather than rely on international restatements, because the commissioning note required it. The National Statistics Office has no resolvable website: nso.gov.er did not resolve at the DNS layer, and the UN Statistics Division's own directory of national statistical office websites has no entry for Eritrea. We did reach two Eritrean official sources — the Ministry of Information's shabait.com, which publishes a country profile and a 1993 referendum retrospective, and the National Statistics Office's own survey reports of 2003 and 2013, which are the origin of the 'No population census has ever been carried out in Eritrea' quotation used twice in this piece. The Ministry of National Development report cited as (MND, 2010) inside the 2010 survey we could not obtain.
- The comparison between 669,412 Eritreans abroad at end-2024 and a national estimate built without a migration term is ours, drawn across two cited sources. We do not claim the projection is wrong by that amount, or by any amount. We claim only that the method as filed contains no term capable of absorbing it.
- On immunisation we describe the divergence between administrative and estimated coverage and quote WHO and UNICEF's own note about an independent World Population Prospects denominator. We do not assert which denominator is closer to the truth, because no source we opened does.
- Coordinates in the panel come from the GeoNames gazetteer, cross-checked against the coordinate UNESCO publishes for the Asmara property. Population figures attached to gazetteer entries are gazetteer values and rest on no enumeration; we print them as such.
- Sources we could not read, and therefore did not cite: the Humanitarian Data Exchange common operational dataset for Eritrean subnational population, 403; the ReliefWeb record for the 2010 survey, 403; the DHS Program's Eritrea country page, 403; the IHSN catalogue record for Ethiopia's 1984 census, robots-blocked; UNHCR's Eritrea country pages, which returned 403 or omitted the figures entirely, which is why the API and the Ethiopia operational portal are cited instead; UNICEF's Eritrea Annual Report 2024, 403; UNDP's Eritrea Annual Report 2023, which contained no machine-readable text; the UNESCO document list for Asmara, 403, and the ICOMOS evaluation behind it; the UN Archives Geneva record for UNOVER, which returned only a browser verification screen; the UN Digital Library search interface, robots-blocked; and the full text of the Special Rapporteur's report A/HRC/56/24, of which only the summary page was readable.
- Things we could not verify: the total population of Eritrea, by any route; the population of any of the six zobas, none of which has an enumerated figure we could find; the population of Asmara, for which UN Data gives 928,800 and GeoNames 563,930 without either citing a count; the age structure of the country, every published figure being a medium-variant projection; what the 2002 population quick count actually consisted of, no report or methodology for it being obtainable; what the Ministry of Local Government's population count consisted of, or why two official reports nine years apart both give about 3.2 million; whether the EPHS 2025 report has been printed or disseminated, UNDP having said in May 2026 that it was scheduled for 2026; how many countries carry a dash in all four rounds of the UN census register, our reading of that table being partial; how many people visit Eritrea in a year, the World Bank arrivals series being empty since 2016; whether a census is funded or scheduled for the 2030 round, the register recording only that one is expected; and the current cost and success rate of the travel permit required beyond twenty-five kilometres from Asmara, which the FCDO describes only as taking several days and not always succeeding.
AI is used to draft this text, and a human checks it. Every figure, date and quotation above was verified by Tony Hall against the source cited for it in the reference list; where sources disagree, the disagreement is shown rather than resolved, and where a claim could not be confirmed it is either marked unverified or left out. The illustrations are AI-created and labelled as such — no photograph on this site is presented as a record of a real place. Full disclosure · Editorial policy.
Access details change — recheck before you travel. Found an error? Tell us — corrections are dated and shown in place.
AI illustration


