How long does an ADU take to approve?

The 60-day rule, under the current numbering

Older guides still cite Government Code § 65852.2 for the ministerial ADU shot clock. That section was repealed.

SB 477 (Stats. 2024, ch. 7), an urgency measure effective 25 March 2024, recodified State ADU Law into Chapter 13 of Division 1 of Title 7, sections 66310 through 66342.

The 60-day decision now lives at § 66317 for accessory dwelling units and at § 66335 for junior accessory dwelling units.

The live text of § 66317(a)(3), as amended by SB 543 (Stats. 2025, ch. 520, effective 1 January 2026), is: the permitting agency “shall either approve or deny the application to create or serve an accessory dwelling unit within 60 days from the date the permitting agency receives a completed application if there is an existing single-family or multifamily dwelling on the lot.” If the local agency has not approved or denied the completed application within 60 days, “the application shall be deemed approved.” If the applicant requests a delay, the 60 days are tolled.

If the ADU application is submitted with an application to create a new primary dwelling, the agency may delay the ADU decision until it decides the primary dwelling, but the ADU application remains ministerial.

SB 543 also wrote a completeness clock that the older § 65852.2 text did not carry in this form.

Section 66317(a)(2) requires a written completeness determination within 15 business days, a list of incomplete items, a limited resubmittal, and a deemed-complete consequence if the agency misses that determination.

An appeal of incompleteness or denial gets a final written determination within 60 business days (§ 66317(d)). Those are business days. The 60-day decision is not labeled as business days.

We measure calendar days on the published clock and do not convert.

The junior-unit counterpart is § 66335, also amended by SB 543: ministerial review, the same 15-business-day completeness machine, and 60 days from a completed application if there is an existing single-family dwelling on the lot.

Coastal development permits are no longer outside that 60-day idea. Section 66329, as amended by AB 462 (Stats. 2025, ch. 491, effective 10 October 2025), requires a local government with a certified local coastal program to approve or deny a coastal development permit for an ADU within 60 days of a completed application, concurrently with the § 66317 decision, without a public hearing, and not subject to appeal under Public Resources Code § 30603.

Where the local government has no certified LCP, the Coastal Commission has the same 60-day duty, with a deemed-approved consequence if it misses.

A narrower 30-day decision still sits outside Chapter 13. Government Code § 65852.27(b) — AB 1332, not recodified into §§ 66310–66342 — requires a local agency to approve or deny a complete application for a detached ADU within 30 days if the application uses a plan preapproved in the current triennial California Building Standards Code cycle, or a plan identical to one the agency already approved in that cycle.

That is a statutory clock, not a measured median. What catalogs actually exist is a different study: pre-approved ADU plans.

Why the contractor harvests cannot answer this question

The HyreADU desk already holds issued-permit extracts for Los Angeles, San Francisco, Sacramento, San José, unincorporated Marin and San Diego, used to name the contractor of record.

Every nested permit object in those files has an issued field. None has an application, filed, submitted or received date. Duration is not a subtraction you can perform on a single date.

The live portals were therefore re-opened on 2026-09-05 for the columns the harvests dropped. LADBS’s current issued feed (pi9x-tg5x) carries submitted_date and issue_date.

DataSF’s building-permit extract (i98e-djp9) carries filed_date and issued_date. Marin’s building-permit extract (mkbn-caye) carries received_date and issued_date.

Sacramento’s issued-permit FeatureServer carries Status_Date and an Application field that is a permit number (RES-2600002), not a date. San José’s CKAN active and expired building-permit resources carry ISSUEDATE and FINALDATE. Two dates or no duration. No imputation.

Los Angeles: 162 days, and a city that is not one counter

The current LADBS issued extract — “Building and Safety — Building Permits Issued from 2020 to Present (N),” resource pi9x-tg5x on data.lacity.org — is the first Los Angeles shape this desk has that carries both submitted_date and issue_date. The legacy contractor extract (xnhu-aczu) does not. We do not mix them.

Filter, labeled as a filter: use_desc contains “ACCESSORY DWELLING” (the department’s own use classification; 36,006 rows in the issued feed), both dates present (35,224; 782 ADU-use rows had no submitted date and are out), Bldg-Demolition dropped (7 rows), negative durations dropped (0).

Issued dates run 2020-01-02 to 2026-08-29. The feed’s rowsUpdatedAt on retrieval was 2026-08-31; the maximum refresh_time on the ADU rows was 2026-08-30.

The headline Los Angeles median excludes work descriptions matching a supplemental-permit keyword. A supplemental is a change to an already-issued file; treating it as a new approval would mix one-day revisions with first issuances.

That cut is 27,188 permits, median 162 calendar days, IQR 79–301. One in five (19%) issued within 60 calendar days of submit; 19% took more than a year.

The mean, labeled as a mean and not the headline, is 236.9 days — seventy-plus days above the median, which is the skew.

Unique primary addresses, taking the earliest non-supplemental submission at each address, are almost the same clock: 169 days, n = 26,005.

Including supplementals is the number a careless row-count would ship: 119 days, n = 35,217, because the counter-plan-check path is full of short files. We report it as a sensitivity. We do not lead with it.

San Francisco: 445 days, and almost a third of the ADU flag has not issued

DataSF’s Building Permits extract (i98e-djp9) is the rare California feed that both flags accessory dwelling units itself (adu = Y) and publishes filed_date and issued_date.

DBI’s data dictionary states that the file is built from permit filing, issuance and completion dates in the Permit Tracking System. We take the city’s flag rather than a text match.

A bare “ADU” in a description is how you hit GRADUATE; we do not do that here.

On 2026-09-05 the extract held 3,268 rows with adu = Y. 1,713 unique permit numbers had both a filed date and an issued date.

Median 445 calendar days, IQR 276–707, issued 2015-06-26 to 2026-09-03. Data as of 2026-09-04.

Same-address duplicates are collapsed by permit number; using only rows with primary_address_flag = Y produces the same n and the same median.

Dropping issued permits later marked canceled or suspend moves the median by two days (443, n = 1,673). We keep the unique-permit set.

Almost none of this clock fits inside 60 calendar days of filing: 0.8% (13 of 1,713). 62% took more than a year. The mean, labeled as a mean, is 549.7 days.

From 2018 the median is 469 (n = 1,589); for files issued 2024 through the retrieval window it is 392 (n = 400, IQR 194–866.2).

A recent-year median below the all-years median is not a finding that the counter has become a 60-day counter.

The other number that belongs above the fold in this city is the one that never enters the median. 952 ADU-flagged records were filed and have not issued.

Status counts on all 3,268 ADU-flagged rows include filed (508), canceled (264), withdrawn (201), approved-but-not-in-the-issued-cut (73) and triage (29). Those files are the right tail a completed-only median cannot see.

If you are holding a filed ADU application in San Francisco, the issued median is the history of the files that made it through, not a queue position.

Permit type on the unique issued set is almost entirely “additions alterations or repairs” (n = 1,641, median 448). New-construction wood frame is 58 files, median 405.5.

Over-the-counter alterations are 13 files and are the only San Francisco cell whose median (33 days) sits near a month; n = 13 is not a path you can count on.

Unincorporated Marin: 174.5 days on a small, description-matched set

Marin County’s building-permit extract (mkbn-caye on data.marincounty.gov) covers permits “processed by, and under the jurisdiction of, Marin County” — unincorporated county land. The ten incorporated towns run their own counters. city_town is a postal city and is not treated as a municipality here.

Unlike Los Angeles and San Francisco, Marin does not classify the permit as an ADU.

We reused the contractor-harvest description test: a word-boundary match on accessory dwelling, JADU, granny flat, second unit, garage conversion or companion unit, after a Socrata like pull, then dropped trade-only rows (solar, reroof, water heater, panel, and the rest) unless a builder verb also appeared. 494 rows had both received_date and issued_date and survived the like-pull; 12 failed the trade-only test; 482 remained.

From 2018 — the first full year of the modern ADU statute — n = 472, median 174.5 calendar days, IQR 105–308.2. Unique addresses are essentially the same file (n = 475, median 175).

Issued dates on the kept set run 2014-04-10 to 2026-09-02; the 2018-onwards cut is the headline because a 2014 n = 3 cell is not a clock.

12% issued within 60 calendar days of receipt; 18% took more than a year. 32 of the 2018-onwards rows issued the same calendar day they were received.

Same-day issuance is in the extract; it is not a typical Marin ADU. Year cells from 2018 through 2026 each have between 23 and 98 issued files. We do not trend a median across cells that small.

This is the smallest of the three samples, the only one that depends on a description regex rather than a city-coded ADU field, and the only one that is a county unincorporated area rather than a city. It is in the table because both dates exist. It is not Los Angeles with a different name.

What Terner already measured, and why this page is not a restatement

Quinn Underriner, “Comparing ADU Permitting Time Inside and Outside the Coastal Zone,” Terner Center for Housing Innovation, 21 March 2024, is the published comparison of ADU permitting time by Coastal Zone.

We cite it as Terner’s. We do not restate it as a HyreADU coastal finding, and we did not geocode these three extracts against the Coastal Zone boundary.

Terner used HCD Annual Progress Report Tables A and A2 for 2018–2022 in Los Angeles County, Orange County and San Diego County, merged on APN, geocoded, and compared average days from a planning application to a building permit.

Their headline averages: Los Angeles County, 260 days in the Coastal Zone versus 147 days outside it; Orange County, 233 versus 101; San Diego County, no statistically significant difference.

The Los Angeles and Orange County gaps were significant at p < 0.01. Those are county-level averages of an APR application-to-permit clock, including jurisdictions this page never opened.

Terner’s own limitation is the reason this page exists as a complement rather than a copy.

Of roughly 60,000 ADU projects in the APR with a valid permit date, they could identify both an application date and a permit date for about 29,000.

“Applications with usable dates may be different from those without, leading to bias in our timeline estimates.” A robustness check on HCD-cleaned dashboard data showed the same direction and significance with moderately lower averages.

Their endnote 3 cited the 60-day local-agency limit at Government Code § 65852.2, the numbering in force when they wrote; that section has since been recodified to § 66317, as set out above.

HyreADU analysis: Terner answered a geographic question this extract is not built to answer — Coastal Zone versus not — with a source this extract is not (the APR), a statistic this page refuses as a headline (the mean), and a window that ends in 2022.

We answered a municipal-counter question they did not publish as medians: submitted-to-issued in the City of Los Angeles after 2020, filed-to-issued in San Francisco, received-to-issued in unincorporated Marin.

A 162-day Los Angeles city median on the building-permit clock and a 147-day Los Angeles County average outside the Coastal Zone on the APR clock are not the same number, and stitching them into one “Los Angeles takes X days” sentence would be the error this desk is here to avoid.

The Coastal Zone remains a reason a coastal lot can sit outside the inland median; AB 462’s concurrent 60-day CDP clock is too new, as of this retrieval, to have rewritten the history in these extracts.

The question everybody asks, and the sentence we will not print

Search results for this question still offer a band — most often three to six months, sometimes 60 days because the statute says 60 days, sometimes a year because a neighbor’s garage conversion took a year.

Those sentences rarely name a city, a year, a sample size or a start date. Sixty days is the statutory decision from a completed application.

A year is a point on the San Francisco IQR. Three to six months is not a measurement we can attach to a method.

HyreADU analysis: the useful answer is a median, an IQR, an n, a window and a list of the cities that would not support even that. Where we cannot compute duration we say so.

Inventing a statewide “ADU permit takes X months” from three municipal extracts and a Terner county-average blog would be the same sin as the landing pages, with a bibliography attached.