How long an Austin building permit actually takes
What this measures. The number of calendar days between the date an application was filed and the date the permit was issued, for every new single-family permit in the city's own records. It is not an estimate and it is not our experience — it is a count.
What it does not measure. Design, engineering and the time it takes to assemble a submittal all happen before the clock starts. Construction happens after it stops.
| Permits issued in | Permits | Median days | 25th pct | 75th pct | 90th pct |
|---|---|---|---|---|---|
| 2014 | 2,623 | 16 | 11 | 36 | 77 |
| 2015 | 2,663 | 27 | 14 | 52 | 103 |
| 2016 | 3,571 | 19 | 13 | 38 | 84 |
| 2017 | 3,751 | 20 | 11 | 46 | 98 |
| 2018 | 3,605 | 27 | 16 | 54 | 102 |
| 2019 | 3,884 | 22 | 10 | 51 | 118 |
| 2020 | 3,608 | 28 | 9 | 61 | 138 |
| 2021 | 3,718 | 38 | 20 | 73 | 129 |
| 2022 | 2,898 | 51 | 16 | 100 | 204 |
| 2023 | 1,565 | 71 | 25 | 198 | 374 |
| 2024 | 1,788 | 51 | 24 | 93 | 259 |
| 2025 | 1,978 | 59 | 27 | 119 | 223 |
| 2026 (through Sep 3) | 1,507 | 48 | 25 | 91 | 205 |
| Last 365 days (Sep 4 2025 – Sep 3 2026) | 1,983 | 54 | 27 | 109 | 234 |
The series starts in 2014 because the city's earlier records do not carry a real application date. In the older rows that field was backfilled with the issue date: all but 5 of the 39,860 permits issued through 1999 show application and issuance on the same day, which is a data artifact, not a same-day permit. The share is still 7.8% across 2007–2010 and falls to 0.1% from 2014 on. Showing the full series would publish tens of thousands of permits as "0 days". The last row is a partial year. The city's data runs through September 3, 2026, so the 2026 row covers part of the year and cannot be read beside the full years above it. Cut 2025 to the same January 1 – September 3 window and its median is 52 days rather than the 59 its full year shows — so most of the apparent improvement is the shorter window, not faster review. The last row is a rolling window, not another year. It covers the 1,983 permits issued in the 365 days ending September 3, 2026, which is a full year of issuances measured to the same cut-off — so it can be read beside the complete years above in a way the part-year row cannot. It overlaps them: every permit in it is already counted in the 2025 or 2026 row, so the rows do not sum and the window is not an addition to the series. Against the 59-day median for the whole of 2025, the trailing year runs 54 days across a near-identical 1,983 permits.
Why these are grouped by the year the permit issued, not the year it was applied for. The city's dataset contains issued permits only. An application filed last month that has not yet been approved is not in it. Grouping by application year would therefore drop exactly the slowest cases from the most recent years and report a median that is too good. Grouping by issue year avoids that: every permit in a given year's group is, by definition, already in the data.
The trade-off is that a permit issued in January may have been applied for two years earlier, so each year's figure reflects a mix of application vintages. In the 2026 cohort that mix is 1,092 permits applied for in 2026, 392 applied for in 2025, and 23 older than that. We disclose the mix rather than adjust for it, because adjusting requires assumptions the data cannot support.
Excluded from the 2026 figures: 16 of 1,523 permits — 0 with no application date recorded, 0 with no issue date, 0 recorded as issued before they were applied for, and 16 with a gap of more than 1500 days, which indicates a reissue against an old application rather than a slow review. That last exclusion is not free: every one of those permits sits in the slow tail, so removing them pulls the 90th percentile down from 223 days to the 205 days shown above. Both numbers are here because a page about slowness should not quietly trim its slowest cases.
A permit is not a project, and we checked what that costs. Builders file in batches: several houses on one application date, issued together. In the 2026 cohort the largest single batch is 14 permits, and 722 of 1,507 permits share an application-and-issue date pair with at least one other. Drop every batched permit and 785 remain.
The median does not move: 48 days across all permits, 48 days across the unbatched ones. The tail does: the 90th percentile falls from 205 days to 182. Read the 90th percentile as a range of roughly 182–205 days rather than as a single figure. These are 1,475 distinct addresses across 1,507 permits, so they are individual houses rather than one development counted repeatedly — which is exactly what this check is for, and it is why we publish no equivalent figure for apartment buildings. In the same dataset, a handful of large multifamily projects each generate dozens of permit rows sharing one application date, and any median computed from them describes almost no real projects.
Why this number is larger than the ones you may have seen. City-wide permit-speed rankings quote a single median across every permit type a city issues. Austin issues far more small permits — mechanical, electrical, plumbing, water heaters — than new houses, and many of those turn around in days, so a pooled median is pulled well below what a new house actually waits. One widely cited index reports a 22-day median for Austin from 514,000+ permits drawn from the same city open-data portal used here, and ranks Austin among the fastest cities in the country.
That number is not wrong; it answers a different question. Restricted to new single-family houses, the same city records, measured the same way — application filed to permit issued — give 48 days. Both figures come from Austin's own data. The difference is entirely which permits are counted, which is why we state the permit class (R- 101) on every figure on this page.
We deliberately do not compare against plan-review timelines quoted in business days, or against figures drawn from a vendor's own project book. Those measure different intervals over different populations, and setting them beside this one would produce a comparison that looks rigorous and is not.
| ZIP | Permits | Median days | 90th pct |
|---|---|---|---|
| 78652 | 44 | 15 | 18 |
| 78748 | 28 | 15 | 140 |
| 78617 | 91 | 20 | 50 |
| 78747 | 180 | 24.5 | 84 |
| 78744 | 291 | 33 | 140 |
| 78653 | 92 | 35.5 | 67 |
| 78754 | 104 | 38.5 | 62 |
| 78724 | 34 | 44.5 | 75 |
| 78757 | 60 | 63 | 146 |
| 78745 | 92 | 66.5 | 210 |
| 78731 | 41 | 82 | 257 |
| 78723 | 50 | 86 | 462 |
| 78704 | 74 | 91 | 224 |
| 78702 | 66 | 94 | 363 |
| 78703 | 27 | 133 | 272 |
| 78721 | 41 | 180 | 360 |
This is a map of what gets built where, not of how fast each ZIP is reviewed. 78652 at 15 days and 78721 at 180 days are not two speeds of review — they are two kinds of project. A ZIP dominated by repeat-plan production housing on platted subdivision lots clears review quickly because the same plan has been reviewed before; a ZIP of one-off infill on constrained lots does not. Nothing here says a permit would move faster if the lot were in a different ZIP. The table shows the 16 ZIP codes with at least 25 permits in 2026 — 1,315 of the 1,507 permits in the cohort, with 192 in thinner ZIPs not shown. The threshold only removes the thinnest cells; it does not make the rows that survive it stable. 78703 clears it with 27 permits and still runs from 29 days at its fastest to 416 at its slowest, with a median of 133 and a 90th percentile of 272. Read every row as a rough shape, not a number.
Using these figures. Please do — they are computed from public records and we would rather they were quoted correctly than reconstructed badly. Quote or screenshot anything on this page with attribution to KiliGray Homes and a link back to it. No permission needed and nothing to sign.
Cite it as: KiliGray Homes, “How long an Austin building permit actually takes,” computed from the City of Austin Issued Construction Permits dataset (3syk-w9eu), permits issued between January 1 and September 3, 2026, retrieved 2026-09-06. The underlying dataset is public and linked above, so every figure here can be recomputed independently — and if you recompute one and get a different answer, we want to know.
Reporters and researchers: we will run a cut of this for you — by ZIP, by year, by council district — and say plainly when the sample is too thin to support the question. Contact details are on our contact page.
Want to know what your lot is actually looking at?
These are city-wide medians. What your project takes depends on the lot — watershed, trees, whether it needs a variance. We pull permits in Austin constantly and can tell you which bucket yours is likely to land in.
Ask about your lotWe are homebuilders, not the City of Austin. This page explains how we read the code in practice — it is not legal advice and it is not an official determination. Always confirm with Austin Development Services before you rely on it for a project. The authoritative text is the Austin Land Development Code; permitting questions go to Austin Development Services Department.