Census vs Estimate: Why Population Figures Disagree

Look up Pakistan's population and you may find 241 million in one source and 248 million in another, both for 2023. Neither is a typo. One is a census count and the other is a UN estimate, and they measure slightly different things on different dates. This guide explains where population figures come from, why they disagree, and what the number on a Beelad country page means.

Census counts and estimates

A census tries to count every person in a country at one reference moment. China's 2020 census, for example, used 00:00 on 1 November 2020 and counted 1,411.78 million people. The US census of 2020 counted 331,449,281 residents as of 1 April 2020. A census is the most complete source a country has, but it is expensive, so most countries run one only every five or ten years. UN guidance recommends a census at least every 10 years.

An estimate fills the years in between. Statisticians start from the last census and add births, subtract deaths and adjust for migration, year by year. This is called the cohort-component method, and it is how national offices and the UN produce figures for years without a census. The further an estimate is from its last census, the more it depends on assumptions.

The UN World Population Prospects (WPP), published by the Population Division of UN DESA, applies one method to every country. The 2024 revision covers 237 countries and areas and draws on 1,910 national censuses. Census counts are "adjusted, where necessary, for coverage gaps, deficiencies in age reporting, and over- or under-enumeration". The published figure for 1 July of a year is the average of the 1 January estimates on either side.

The same revision also looks forward. It puts the world at 8.2 billion people in 2024 and projects a peak of around 10.3 billion in the mid-2080s, falling slightly to 10.2 billion by 2100. These projections are scenarios built on assumptions about fertility, mortality and migration, so treat them as a different kind of number from both census counts and estimates of the past.

De facto and de jure: two ways to count

Before a census begins, a country has to decide where each person belongs.

• De facto: people are counted where they are physically present at the census moment. Visitors are included and residents who are abroad are not.
• De jure: people are counted at their place of usual residence, wherever they happen to be that night. Residents away on a trip are included and short-term visitors are not.

Neither rule is about citizenship. A common mistake is to treat "de jure" as "citizens only". In the United Arab Emirates, foreign workers who live there are usual residents, so they are included in a de jure count and a de facto count alike. Emirati nationals are roughly 11% of the population, according to commonly cited figures, but the gap between residents and nationals is a separate question from de jure versus de facto. Countries with many tourists or many seasonal workers abroad are where the two rules give noticeably different totals.

Census and UN figures for the 10 largest countries

The table sets the UN estimate shown on Beelad beside each country's latest census count. The last column is not an error rate: the dates differ, so most of the gap is population growth since the census.

A few rows need a note. Bangladesh's 2022 census first reported 165.1 million and was raised to 169.8 million after a post-enumeration check. Russia's 2021 total includes Crimea and Sevastopol, annexed in 2014, so it is not directly comparable with the UN figure. China's census covers the mainland only and excludes Hong Kong, Macao and Taiwan. Beelad lists Taiwan as a separate entry; Hong Kong and Macao are not in its list. The US count covers the 50 states and the District of Columbia.

Countries counting from old censuses

The UN estimate is only as good as the data under it, and some large countries have not counted for a long time.

• Nigeria last counted on 21 March 2006: 140,431,790 people. The UN estimate for mid-2023 is 227.9 million, which implies growth of about 2.8% a year for 17 years. All of that rests on surveys and modeling. See the Nigeria page.
• India has not published a census count since 2011, when it recorded 1,210,854,977 people. The 2027 census began its house-listing phase on 1 April 2026, and its population count uses a reference date of 00:00 on 1 March 2027 (1 October 2026 for snow-bound areas of Ladakh, Jammu and Kashmir, Himachal Pradesh and Uttarakhand).
• Ethiopia's last census was in 2007; the next one has been postponed several times since its original 2017 date.

India and China: a crossover set by estimates

When India passed China as the most populous country, no one counted it happening, because neither country held a census at the time. A UN projection published in April 2023 put the crossover around that month. The WPP 2024 revision, with newer data, puts India slightly ahead already on 1 July 2022, by 243,643 people, and 15.5 million ahead a year later. Dates like this move with each revision.

What Beelad shows

Every Beelad country page uses the same source: the UN WPP 2024 estimate for 1 July 2023. The 196 countries in Beelad's list add up to 8,069,353,071, which is 22.4 million short of the UN world total of 8,091,734,930. The difference is the territories and areas that the UN counts separately and that are not in Beelad's list of countries. For the smallest populations on the list, see the smallest countries in the world; continent hubs such as Asia let you sort countries by population.

Quick checklist

• Check the date and the type of figure: census count, national estimate or UN estimate.
• Compare countries using one source for the same date, such as the UN WPP figures for 1 July.
• Use the census for legal and administrative purposes inside a country.
• "De jure" means usual residents, not citizens. Foreign residents are included.
• Treat figures for countries with old censuses, such as Nigeria (2006), Ethiopia (2007) and DR Congo (1984), as estimates with wide uncertainty.