Two official population figures can disagree without either being a reporting mistake. A decennial count records population on a census date; census estimates use that starting point and later evidence to describe population between censuses.
The difference usually rests on date, method, and revision history. We keep those pieces attached to the number before deciding what a mismatch means.
Start with what each figure measures, because a shared Census Bureau source doesn't make the products interchangeable.
Decennial counts and census estimates answer different questions
A count describes one census date
The 2020 Census counted people according to their usual residence on April 1, 2020. Enumeration means counting people, with the goal of including everyone in the population covered.
Collection continued beyond that day, but the reference date stayed fixed. A response collected later still belonged to the April 1 count.
That distinction matters when an older count appears in a newer story. The figure can accurately describe the census date while giving an outdated picture of a community's present size.
An estimate describes a later population
Annual estimates account for changes after the census. Their standard reference date is July 1. They describe the estimated population on that date, rather than an average across the year.
The basic comparison looks like this:
| Feature | Decennial count | Annual population estimate |
|---|---|---|
| Reference date | Census Day, April 1 | Usually July 1 |
| Main method | Nationwide enumeration | Census-based starting population plus later evidence |
| Timing | Every 10 years | Annual releases |
| Earlier figures | Fixed census-date reference | Prior estimates may be revised |
The dates alone can explain part of a difference. April 1, 2020 and July 1, 2020 are three months apart.
Our guidance on checking population claims follows the same rule: match the figure to the claim before interpreting the gap.
Population estimates start with a base
The base population is the starting number
The base population is the population used as the starting point for calculating later estimates. It is derived from the census, but it doesn't have to match every published decennial total.
The Census Bureau can incorporate corrections, geographic changes, and other adjustments into that base. That gives later estimates a different foundation than a simple copy of the original count.
An adjustment to the estimates base doesn't automatically rewrite the published decennial count. Those records have different purposes. We need to identify which one a table contains before treating two April 1 figures as contradictory.
Adjustments can change the foundation
Vintage 2025 uses a census-derived base that incorporates the Count Question Resolution and Post-Census Group Quarters Review programs. These address certain problems involving geographic placement, housing units, and facilities such as dormitories or correctional institutions.
It also adjusts county-level base populations for children ages 0 through 10, drawing on research about undercounting in the 2020 Census.
These details matter because a later estimate can differ even before subsequent births, deaths, or moves enter the calculation. The starting number itself may have changed.
The Bureau's Population Estimates Program publishes the estimates and supporting documentation. The base is part of the method, not a detail we can strip away.
Births, deaths, and migration move the total
The components of change are the events that increase or decrease population. The main components are births, deaths, and migration.
Births and deaths explain natural change
Births add residents. Deaths remove residents. The difference between them is called natural increase when births exceed deaths, or natural decrease when deaths exceed births.
For Vintage 2025, the Bureau uses National Center for Health Statistics data for births and deaths. These records update the population picture without another nationwide enumeration.
The Bureau's population estimates guidance explains how current demographic information is used to calculate change after the census. The framework is straightforward, although producing the inputs takes time.
Migration explains movement across boundaries
Net migration is the number moving into an area minus the number moving out.
Domestic migration involves moves within the United States. International migration involves movement between the United States and other countries. A move between states changes those states' populations without adding a resident to the national total.
That is where a population headline can outrun its evidence. A growing total doesn't establish that migration caused the growth. Births and deaths also matter.
We separate the total change from its explanation. The components, rather than the headline number alone, show which measured forces pushed population up or down. Smaller-area estimates may also use housing-unit information to distribute population within counties.
Why an earlier estimate can change
A vintage identifies a revised series
A vintage is an annual edition of the estimates series. Vintage 2025 includes estimates through July 1, 2025, even though its releases appeared in 2026.
Each new vintage can revise earlier estimates as well as add another year. Updated records, improved migration measures, and changes to methods can alter the earlier figures.
A revised estimate for 2024 doesn't mean the population changed after July 1, 2024. It means the Bureau's estimate of that earlier population changed.
For year-to-year growth, use both endpoints from the same vintage. Mixing releases can fold a revision into what looks like population change.
An older article may have accurately reported the release available at publication. A later estimate can differ without proving that the original report copied the number incorrectly.
Estimates between two censuses have another anchor
Postcensal estimates describe years after the latest census, using that census as the starting point.
Intercensal estimates describe years between two completed censuses. They can use the later census to adjust the earlier series so it connects the two census-based endpoints.
The Bureau's explanation of 2010 to 2020 intercensal methodology describes this process and its limits.
The later census adds information that wasn't available when the original annual estimates were produced. That can reveal differences that accumulated across the decade.
Both counts and estimates have limits
A census seeks a complete count, but complete coverage is a goal rather than a guarantee. Some people are missed. Some are counted more than once. Others are assigned to the wrong location.
A coverage error means the count misses or includes people incorrectly. These problems can affect groups and places differently, even when the overall total looks close.
Estimates carry uncertainty of their own. They depend on the starting population, the completeness of records, and the methods used to measure movement. Migration is harder to track than a registered birth or death.
A difference between a count and an estimate can therefore reflect several things at once: actual change, a revised base, imperfect coverage, or changed methods.
We shouldn't turn that difference into a broad accusation without identifying the cause. Nor should we assume that an official number answers every question attached to it.
An estimate displayed to the nearest person is still an estimate. Display precision tells you how the number is printed. It doesn't establish that every resident was individually counted.
ACS estimates describe a different kind of evidence
The American Community Survey, or ACS, is another Census Bureau product. It measures characteristics such as income, education, employment, and housing conditions.
It uses sampling, which means collecting information from a selected portion of the population to estimate characteristics of the larger group. That differs from the decennial census's effort to enumerate everyone.
It also differs from the Population Estimates Program's census-derived base and demographic accounting framework. The two programs aren't interchangeable because both publish estimates.
An ACS one-year estimate covers 12 months. An ACS five-year estimate covers 60 months. Neither is a July 1 population snapshot.
That difference in timing matters. A five-year estimate combines information collected across a longer period, so its label doesn't mean the result describes only the final year.
ACS figures also come with margins of error, which describe sampling uncertainty. Those margins don't capture every possible source of error.
For comparisons, keep one-year ACS estimates with one-year estimates and five-year estimates with five-year estimates. We choose the product that measures the question, rather than the number that is easiest to find.
How to compare two population figures fairly
A useful comparison preserves the place, date, people, and release behind each figure. Our guidance on tracing original data sources applies here because summaries often leave those details behind.
- Identify the geography and population covered. A city, county, and metropolitan area have different boundaries. Total residents include children and aren't a count of eligible voters.
- Match the reference dates and data products. An April 1 census count, a July 1 population estimate, and a five-year ACS estimate measure different things.
- Keep the vintage and table columns attached. For annual growth, use two dates within the same estimates series.
Boundary changes need attention too. Annexation can increase a city's reported population because its territory expanded. That doesn't mean the same area gained that many residents.
For national resident totals, NST-EST2025-POP contains the Vintage 2025 population figures. The Bureau's national population tables provide totals and components of change.
Percentage growth uses the earlier population as its denominator: subtract the earlier total from the later total, divide by the earlier total, then multiply by 100. Use full values before rounding.
Finally, distinguish the release date from the reference date. Publication in 2026 doesn't turn a July 2025 estimate into a 2026 population figure.
Key takeaways
- A decennial count describes population on Census Day. Annual estimates describe population on later reference dates.
- Differences can reflect births, deaths, migration, base adjustments, boundaries, and revised methods. A total alone doesn't identify the cause.
- A fair comparison uses compatible geography, population definitions, dates, products, and vintages. ACS period estimates require separate treatment.
These distinctions keep a valid number from acquiring a claim the underlying data don't support.
Conclusion
A mismatch between a decennial count and an annual estimate needs an explanation before it becomes a verdict. The strongest explanation keeps the starting population, reference date, and revision history visible.
We read the number alongside its method. You can then distinguish actual population change from a changed estimate of that population.
An official figure is useful evidence. Its meaning depends on what was measured, when it was measured, and which version of the record you are reading.