ADU legalization statistics

The six levers, verified state by state

Read the key before the cells. A filled cell means the provision was verified present in the statute we retrieved and is quoted in the table. A dashed cell means we did not retrieve it in this edition, and treating that as an absent provision would be exactly the misreading the page exists to prevent.

HyreADU coding of state statutes retrieved 2026-09-05 from each state’s own legislature site or its official statutory publisher: leginfo.legislature.ca.gov (California), app.leg.wa.gov (Washington), oregon.public.law (Oregon), legislature.maine.gov (Maine). Verified counts: California 6 of 6, Washington 4 of 6, Maine 3 of 6, Oregon 2 of 6.

The covenant override is the lever most often forgotten, and we verified it in California alone. Civil Code section 714.3(a) makes any covenant that “either effectively prohibits or unreasonably restricts the construction or use of an accessory dwelling unit or junior accessory dwelling unit … void and unenforceable”, subject to a reasonable-restrictions proviso at subdivision (b) that expressly excludes fees or financial requirements.

In a state with extensive common interest developments, it is the provision that decides whether preemption of the city achieves anything at all — a homeowners association can otherwise reproduce every requirement the statute removed.

Whether the other three states have an equivalent we did not establish.

The same count, drawn

One bar per statute read. The bar is a count of what we verified, and a shorter bar has two possible explanations that cannot be told apart from the chart: a shorter statute, or a shallower retrieval.

HyreADU coding, 2026-09-05. The “not retrieved” column is published so the shape of the matrix is explained rather than merely apologized for.

The five-provision fingerprint across seven statutes

A wider and shallower cut, from the model ordinance study: five provisions coded across seven state ADU statutes, each read directly. Four of the five recur almost everywhere. The fifth does not.

HyreADU coding of seven state ADU statutes, 2026-09-05. “Variant” means the state restricts the requirement in a narrower form rather than prohibiting it outright — Washington caps parking rather than barring it, and Colorado’s owner-occupancy and parking provisions are qualified. Every cell is reproducible from the statutory text cited on the model ordinance study.

The comparison that could not be run, stated plainly. The model ordinance study was commissioned to compare adopted statutes against AARP’s published model ordinance.

AARP’s own resource page names the publication — “Accessory Dwelling Units: Model State Act and Local Ordinance”, described there as a 56-page guide — but its operative text could not be retrieved.

No claim about what the model says appears on that page or on this one, and no statute is described as derived from it. A text comparison against a document nobody on this desk has read would be a fabrication.

The clause nobody copies: who the statute binds

Who the statute binds is the finding of the fingerprint work, and it is worth more than any of the substantive rows. The rules about parking, owner-occupancy and approval look like a shared template across seven states.

The applicability clause looks like seven separate political negotiations, because that is what it is — it decides which local governments lose authority, so a legislature writes it itself.

Applicability provisions in seven state ADU statutes, retrieved 2026-09-05. Only 3 of the seven — California, Montana and Maine — apply with no population or geographic trigger.

Arizona is the clearest illustration of why this clause is the one that matters. Its ADU statute is strong on every substantive row in the fingerprint — by right, no owner-occupancy, no additional parking — and applies only to municipalities with a population of more than seventy-five thousand.

A statute can be maximally permissive and reach almost none of a state’s jurisdictions. Any count of “states that have legalized ADUs” which does not read this clause is counting the wrong thing.

The concordance, and how much of it we verified

Where each provision used to live and where it lives now. 2 of the 5 entries are verified from the code’s own enactment line — the sentence the Legislative Counsel prints at the foot of the section — and the rest are recorded on weaker evidence, which is stated in the row rather than hidden.

HyreADU concordance, compiled 2026-09-05 and re-verified 2026-09-07. The § 66311.5 and § 66311.7 rows carry the renumbering sentence from the code itself; the § 66327 row does not and says so.

Why we publish the weak row rather than dropping it. A concordance that quietly omits the entry it is least sure about is more dangerous than one that flags it, because a reader cannot tell which entries were checked.

The § 66327 row is the one to treat with caution: we established that § 66327 does not appear in the current chapter and that § 66313.5 does, which is consistent with a renumbering but is not the same as reading the renumbering line. The owner-occupancy study carries the full statutory history.

What happened afterwards, in the one state where we can measure it

A legalization statistic is more useful with an outcome beside it — and much more dangerous if the two are presented as cause and effect.

These are HyreADU calculations from HCD’s own Annual Progress Report Table A2, deduplicated so a project reported in successive years is counted once.

They are placed here for context and they establish no causal claim.

HyreADU calculation from HCD Table A2, HCD file last updated 2026-09-04. Permits are unique (jurisdiction, project identifier, building permit date) events; completions are unique certificate-of-occupancy events. Deduplication removed 35,237 permit units, 16.5% of the naïve total — which is why a figure taken straight from the annual rows will be higher than ours.

Do not read this table as an effect of the statutes on the timeline. Interest rates, construction costs, the pandemic, the CalHFA grant program, the sixth housing element cycle and the maturing of a contractor market all move in the same window, and we have not attempted to separate them.

What the table does establish is that the accessory route went from a rounding error to a quarter of everything California reported, over the same period in which the statute was rewritten five times.

The causal question is open, and we would rather say so than draw an arrow.

Every figure on this page, with its class

The full ledger, so the 27-of-42 split can be checked rather than taken on trust. Nothing appears anywhere on this page that is not in this table.

27 HyreADU calculation · 15 Agency published · 0 Trade / survey estimate. The hub gate requires at least one third to be ours; this page is at 64 per cent.