What a building permit count can and cannot tell you
A building permit count tells you how many new housing units local governments authorized in a year. It does not tell you how many were started, finished or occupied, and it leaves out places that require no permit. In 2025 the count was 1,431,616 units for the United States, of which the Census Bureau estimated 9.3% for permit offices that did not report every month.
Authorized is not built
Permits come first in the life of a new home. Some projects are delayed, redesigned or dropped after the permit is issued. The Census Bureau runs separate surveys of housing starts and completions; this site uses only the permit survey, because it is the one published for every county and metro area. A year with many permits signals plans, not finished homes.
Reported and estimated
About 20,000 permit offices are asked to report. When an office misses months, the Census Bureau imputes the missing figures from its earlier reports and from similar offices. The table compares the totals with and without those estimates.
| Year | Units, with estimates | Units, reported only | Estimated share |
|---|---|---|---|
| 2025 | 1,431,616 | 1,298,881 | 9.3% |
| 2024 | 1,478,000 | 1,336,021 | 9.6% |
| 2023 | 1,511,102 | 1,378,639 | 8.8% |
| 2022 | 1,680,368 | 1,582,647 | 5.8% |
| 2021 | 1,736,982 | 1,604,174 | 7.7% |
| 2020 | 1,471,141 | 1,326,384 | 9.8% |
| 2019 | 1,386,048 | 1,314,345 | 5.2% |
| 2018 | 1,328,827 | 1,217,094 | 8.4% |
| 2017 | 1,281,977 | 1,191,918 | 7% |
| 2016 | 1,206,642 | 1,120,816 | 7.1% |
| 2015 | 1,182,582 | 1,099,552 | 7% |
| 2014 | 1,046,363 | 976,353 | 6.7% |
| 2013 | 990,822 | 923,698 | 6.8% |
| 2012 | 829,658 | 772,277 | 6.9% |
| 2011 | 624,061 | 574,796 | 7.9% |
| 2010 | 604,610 | 562,274 | 7% |
| 2009 | 582,963 | 543,408 | 6.8% |
| 2008 | 905,359 | 841,041 | 7.1% |
| 2007 | 1,398,415 | 1,304,617 | 6.7% |
| 2006 | 1,838,903 | 1,713,527 | 6.8% |
| 2005 | 2,155,316 | 2,007,022 | 6.9% |
| 2004 | 2,070,077 | 1,960,131 | 5.3% |
Nationally the estimated part has stayed between 5.2% and 9.8%. It varies far more by state. From 2021 to 2025 it was highest in Maine (31%), Alaska (28%), Vermont (24%), Montana (22%), Hawaii (22%) and lowest in District of Columbia (0%), Delaware (0.1%), Maryland (0.5%), Nebraska (2.6%). Each state page states its own figure.
Places without permits
Where no permit is required, new homes do not appear in the survey. This matters mostly in rural counties. For metro areas the effect shows up as a very low rate per resident: 24 of the 387 metro areas averaged less than one unit per 1,000 residents a year from 2021 to 2025, well under the national 4.7. Those metro areas are listed in the metro table but have no page here.
Units, not buildings
The survey counts both buildings and units. This site reports units throughout, because a household needs a unit. Shares by building size are shares of units: a 200-unit apartment building adds 200 units to the five-plus class.
Same map for every year
Metro area boundaries are redrawn every few years. When they change, the Census Bureau's own metro tables change with them, so its historical series jump. This site applies the July 2023 definitions to every year from 2004, summing each metro area's counties. A metro area's history here can therefore differ from older published tables, but it always describes the same territory.
Reading one area page
Take Austin. Its page opens with the facts that set it apart, here its 38,332 units a year from 2021 to 2025. The column chart below shows each year since 2004, split into single-family houses, two- to four-unit buildings and buildings of five or more units; the tallest column is labeled with its year. Further down, the page compares Austin's building mix with the country, names the permit offices that issued the most units (the City of Austin and the unincorporated parts of the surrounding counties file separately), and lists the metro areas closest to it in population. A CSV file with every yearly figure is linked at the bottom.
Every rank on such a page names its group, for example all 387 metro areas or the metro areas of a similar size, and each measure is ranked within one group only.
Which metro areas have a page
All 387 metro areas are in the metro table and in Versus, but only 227 have a page of their own. The rest were left without one for one of these reasons: an average of fewer than 300 units a year in 2021–2025 (66 metro areas); fewer than one unit per 1,000 residents a year, a sign of places without permits (24); a county that joined the survey in 2015, which would break the series (4); or too few figures that stand out against other metro areas of a similar size (89). Some metro areas meet more than one of these conditions and are counted under each.
What to compare
Use totals to see where the most homes are authorized, rates per 1,000 residents to compare places of different size, and the five-year averages to look past single large projects. Compare years within the same area with care across 2008–2011, when the count fell to a fraction of its boom level. The counting rules page lists every definition, and Versus plots any two areas on one chart.
Sources
US Census Bureau, Building Permits Survey, annual county and state files 2004–2025; OMB metro area delineation of July 2023; US Census Bureau, Population Estimates, Vintage 2025. All figures computed by this site; see the counting rules.