American Community Survey (5-Year Estimates)

The Bureau publishes several ACS products. Birdseye uses the 5-year estimates, which pool "60 months of collected data" and are the only product released for census tracts. The Bureau describes them as the most reliable of the ACS products and the least current, and they are the only ones available for "data for all areas" (When to Use 1-Year or 5-Year Estimates).

We hold annual vintages beginning with the 2010-2014 estimate period and running through the most recent release. Current values are read from the newest vintage; the earlier vintages exist so that change over time can be measured.

Publication lags the present by more than a year. The 2020-2024 5-year estimates were released on 29 January 2026 (release schedule), and each vintage already describes a five-year window that ended before its release date. A current ACS figure is therefore a five-year average centred roughly three years in the past, not a reading of today.

Every figure is native to the census tract, so no smaller-area estimation is involved. The most recent vintage covers roughly 84,400 tracts nationwide.

Census tract boundaries were redrawn for the 2020 census. Vintages published on the earlier boundaries are carried onto current ones before use, so historical comparisons are made on one consistent set of tracts. See how we handle geography.

ACS estimates come from a sample and carry sampling error. The Bureau publishes a margin of error at the 90 percent confidence level alongside every estimate; margins widen as the population being measured shrinks, so a tract with few households, few renters or few workers has a correspondingly imprecise estimate (Understanding Error and Determining Statistical Significance).

Consecutive 5-year vintages overlap. The Bureau's own guidance is "do not compare overlapping datasets" (Comparing ACS Data), because four of any two adjacent vintages' five years are shared. Trends over five and ten years compare vintages whose windows do not overlap at all; three-year trends compare vintages that share two years, and are damped as a result.

Estimates are suppressed where a tract's sample cannot support them, and dollar figures are inflation-adjusted by the Bureau to the final year of each estimate period.

This product uses the Census Bureau Data API but is not endorsed or certified by the Census Bureau, as required by the API terms of service.

Birdseye figures derived from this source are our own estimates. Per Census Bureau guidance, the Bureau is cited as the source of the original data only, and is not responsible for our calculations.