State Constitutions API: 53 Jurisdictions With Article and Section Structure

State constitutions are available as a queryable corpus at corpusType=STATE_CONSTITUTION, covering 53 jurisdictions (the 50 states, Puerto Rico, the Northern Mariana Islands and American Samoa) across more than 13,000 sections, on a monthly refresh cadence. That is a small corpus by section count and an outsized one by consequence, because state constitutions carry tax caps, school funding formulas, ballot-measure procedure, and privacy rights that go past the federal floor. Most legal-data products skip them entirely, or fold them into the statute corpus where nobody can filter for them. This post covers what is in the corpus, how the article and section structure works, and the queries that make it useful.

Disclosure: we build Vaquill AI, a US primary law API.

Reviewed and corrected on 12 October 2026. Corpus totals and per-state counts below come from the live coverage endpoint as of that day, and the endpoint is free, so you can check the math yourself. This is general information, not legal advice.

TL;DR

• 53 jurisdictions, more than 13,000 sections, filterable as corpusType=STATE_CONSTITUTION, on a monthly refresh cadence. The US Constitution is a separate corpus (corpusType=CONSTITUTION, 74 sections, refreshed when it is amended).
• State constitutions are much longer than the federal one. Arkansas alone carries 513 sections, nearly seven times the federal document's 74. The average jurisdiction runs about 245.
• Count differences between states mostly reflect structure. Some states amend by bolting on new sections. Others push the same policy into the statute book. Raw section count is a poor way to compare providers.
• DC, Guam and the US Virgin Islands have no entry here. None has a constitution of its own. DC's charter is the District of Columbia Home Rule Act, a federal statute, and Guam's and the Virgin Islands' organic acts are federal law. All three are served in the statute, regulation and court-rules corpora.
• Change arrives on a ballot calendar. In many states most amendments go to voters in November, and a few states, such as Texas, hold amendment elections in odd-numbered years as well. That is why a monthly refresh fits the data.
• Point in time is a reconstruction. Each citation returns the current text by default. asOf=YYYY-MM-DD on the body endpoint rebuilds an earlier text from the changes we observed, and isBounded marks answers that rest on when capture began. Amendment history and change events sit beside it.

You will find this alongside the rest of our US primary law corpus series, one post per source.

The two state corpora that sit directly under this one are covered in State Regulations API and State Court Rules API.

The corpus most legal-tech products forget

Ask a legal-data vendor what they cover and you get statutes, regulations, and case law. State constitutions rarely make the list. When they do, they are usually stuffed inside the statute corpus with no separate filter, which means a constitutional question comes back buried under statute sections.

That is a strange gap, because the state constitution is the one document that can invalidate everything else in the state. A statute that conflicts with it is void. A regulation that conflicts with it is void twice over. If your product reasons about state law and never touches the state constitution, it is reasoning one level too low.

The gap is also expensive in specific practice areas. Tax, education finance, election law, municipal finance, and state privacy work all sit on constitutional text more than on statutes.

What is in the corpus

Puerto Rico, the Northern Mariana Islands and American Samoa are in scope because each has a constitution, and each is served the same way a state is. A Puerto Rico hit can come back in Spanish: the Article II, section 10 excerpt returned in our 12 October 2026 test did.

The seven largest constitutions by section count in the corpus, as of 12 October 2026:

Compare those to the US Constitution's 74 sections. Arkansas carries almost seven times as much text-bearing structure as the federal document. Across all 53 jurisdictions the average is about 245 sections, more than three times the federal one, and the median is about 217.

Why state constitutions are so much longer

Three reasons, and all three change how you should query them.

They are easier to amend. The federal Constitution needs two thirds of both chambers of Congress and three quarters of the states. In most states the legislature can refer an amendment to the voters, who approve it by majority vote, and a number of states also allow a citizen initiative that skips the legislature. The details differ by state, so read the amendment article of the constitution you are working in. Oregon's Article IV, section 1 is a well-known initiative and referendum provision. A lower bar tends to mean more amendments, and more amendments mean more sections.

They carry policy as well as structure. The federal document mostly allocates power. State constitutions routinely fix tax rates, cap revenue growth, set school funding formulas, create specific agencies, and pin down pension obligations. Colorado's Article X, section 20, the Taxpayer's Bill of Rights, is a revenue and spending cap written into the constitution itself. That is ordinary legislation in most countries and constitutional text in Colorado.

They are amended by accretion. Many states add a new section rather than editing an old one, so the document grows over time instead of being rewritten. That is why a section count tracks amendment history as much as it tracks subject-matter breadth.

A short history: from John Adams to Alabama's 2022 reset

State constitutions are older than the federal one, and some of them have been through a lot.

The Massachusetts Constitution is the oldest written constitution in the world still in active use. A convention met in September 1779, and the drafting went to a subcommittee of John Adams, Samuel Adams and James Bowdoin, who handed the real work to Adams alone. Adams also sent the result to the voters, another innovation, and it was adopted in 1780 by a two-thirds majority of eligible voters, according to an essay by Northeastern law professor Martha Davis in the State Court Report's 50-state series (April 2025). The same essay says its core provisions have survived since 1780 but the document has been amended more than 120 times.

Then there's Alabama, the stress test for everything above about amending by accretion. Its 1901 constitution had collected 978 amendments by 2022 and ran about 420,000 words, according to the Public Affairs Research Council of Alabama (PARCA). All those amendments sat at the back of the document in the order they were passed. PARCA's example is a good one: Section 102 of the original text forbids interracial marriage, and you only find out voters repealed it in 2000 when you reach Amendment 667. In November 2022 voters were asked to approve a recompiled version that reorganized the text, removed racist language, deleted repealed and duplicate provisions and, in PARCA's words, "makes no change in policy or law." PARCA's analysis put the new text at 373,274 words, which it described as still three times more words than the next-longest state constitution. A 2024 proclamation on the governor's site refers to a proposed amendment to the "Constitution of Alabama of 2022", so the recompiled document is the one in force.

The practical lesson is dull and useful. The recompilation rearranged the text into new articles, parts and sections, so a citation from a pre-2022 brief may need translating before it points at the right place today.

In their words

On why state text matters at all, here is Justice William Brennan in a 1977 Harvard Law Review article, State Constitutions and the Protection of Individual Rights: "State constitutions, too, are a font of individual liberties, their protections often extending beyond those required by the Supreme Court's interpretation of federal law." He wrote it as a working justice telling state lawyers to stop treating the federal Bill of Rights as the whole story.

The idea shows up in casual conversation, too. In a December 2021 Hacker News thread, a commenter put the Montana example in two sentences: "Some states have stronger protections in their state constitutions than the federal constitution provides. Montana, for instance, explicitly provides a right to privacy, whereas a 'right to privacy' is at best implicit in the US Constitution." (antognini, Hacker News, December 2021).

And the ballot-box part of this corpus gets tangled in public debate often enough that a commenter in a March 2023 thread stopped to correct it: "Prop 22 was an initiative statute, and as such did not amend the Constitution." (dragonwriter, Hacker News, March 2023). It's an easy mistake. A measure that voters pass isn't automatically a constitutional amendment, so check what the measure actually was before you search the constitution for it.

Why a raw section count is a bad way to compare providers

The Arkansas number is where that goes wrong. If you line up two legal-data vendors and compare "state constitution sections," the bigger number looks like better coverage. It is usually just a different state mix, a different chunking rule, or a state that happens to amend by accretion.

Look at the spread inside our own numbers. Arkansas is 513. The average is about 245. That isn't because Arkansas is covered twice as well as the average jurisdiction. It is because Arkansas has amended its constitution far more, and each amendment lands as new structure.

The comparison that actually means something has four parts:

Our own endpoints are built for that interrogation. GET /us/statutes/coverage is free and returns per-jurisdiction counts for every corpus plus a measuredAt timestamp. It reports what is held, not how recently it was refreshed, so pair it with GET /boards, which is also free and carries lastRetrievedAt and cadence for each tracked source. Where a publisher states its own currency, that rides on each section as currentThrough. A vendor that won't let you audit coverage without a sales call is telling you something. The coverage endpoint reference documents the fields.

DC, Guam and the Virgin Islands have no constitution, and that is not a gap

Three of the 56 jurisdictions in the wider corpus are missing from this one on purpose. The District of Columbia isn't a state, and its organic law is the District of Columbia Home Rule Act, which Congress passed as a federal statute. Guam's and the US Virgin Islands' organic acts are likewise federal law. The closest equivalents to a constitution for all three live in statutory law.

Putting a "DC Constitution" row in a coverage table would be a fabrication dressed as completeness. DC, Guam and the Virgin Islands are served elsewhere: all 56 jurisdictions carry statutes (corpusType=STATE), state administrative regulations (REGULATION) and court rules (STATE_RULES) in our coverage matrix. See state administrative regulations and state court rules.

This is the same reason the constitution count is 53 and not 56. A corpus that reported 56 constitutions would be wrong, and the extra rows would be the tell.

Worked queries

Base URL is https://api.vaquill.ai/api/v1. Auth is Authorization: Bearer vq_key_.... Statutes routes are mounted under /us/statutes/*. Search costs 4 credits per call and the browse call costs 1, per the pricing endpoint on 12 October 2026. Coverage is free. Failed work bills 0. Every call below also runs in the API playground.

1. Search across every state constitution at once

The single most useful query in this corpus is the fifty-state sweep, because that is the query no statute corpus can answer. The multi-state comparison recipe shows the full pattern.

curl -X POST https://api.vaquill.ai/api/v1/us/statutes/search \
  -H "Authorization: Bearer $VAQUILL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "right of privacy",
    "corpusType": "STATE_CONSTITUTION",
    "limit": 25
  }'

Leave state off and you get the cross-jurisdiction picture. The top six hits in our 12 October 2026 run were Mont. Const. art. II, § 10, Alaska Const. art. I, § 22, Fla. Const. art. I, § 23, N.H. Const. pt. 1, art. 2-b, the Northern Mariana Islands constitution's art. I, § 10, and Puerto Rico's art. II, § 10, with more on later pages. Read each section, because a hit shows where the words appear, not what the provision does. Add "state": ["ca", "mt", "ak", "fl"] to narrow it. state and corpusType both accept a single value or a list.

2. Phrase-match a term of art in one state

curl -X POST https://api.vaquill.ai/api/v1/us/statutes/search \
  -H "Authorization: Bearer $VAQUILL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "equal and uniform",
    "corpusType": "STATE_CONSTITUTION",
    "state": "tx",
    "matchType": "phrase",
    "limit": 10
  }'

matchType takes any, all, or phrase. Use phrase when the exact wording is the thing you care about, which in constitutional text is most of the time. In our 12 October 2026 run this returned Tex. Const. art. 8, § 2 and art. 8, § 1 ("Taxation shall be equal and uniform") as its top two hits, with matchQuality of weak, so read the sections before you rely on the ranking.

3. Browse the article and section tree

Constitutions are organized as articles, then sections, sometimes with a further subdivision. Do not guess the tree. Enumerate it.

curl "https://api.vaquill.ai/api/v1/us/statutes/divisions?corpusType=STATE_CONSTITUTION&state=nm" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"

That returns New Mexico's 23 articles in reading order (Article I, "Name and Boundaries", with 3 sections, through Article XXIV), each with its sectionCount, which is what you want backing a navigation UI. Article labels follow each constitution's own style: New Mexico's read "Article IV", while Texas's hits carry "Article 8". Pass one back as article (&article=Article%20IV) and you get that article's sections, each with the actId you fetch it by.

The same article scopes a search, so you can rank within one article instead of the whole constitution:

curl -X POST https://api.vaquill.ai/api/v1/us/statutes/search \
  -H "Authorization: Bearer $VAQUILL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "legislative sessions",
    "corpusType": "STATE_CONSTITUTION",
    "state": "nm",
    "article": "Article IV"
  }'

It works from the other direction too. Every search hit carries a parent object naming its article ({"corpusType": "STATE_CONSTITUTION", "state": "nm", "article": "Article IV"}), and you can pass it back to /divisions to list the article or spread it into a search body to search inside it. The browse hierarchy recipe walks both patterns.

4. Retrieve one section by its id

curl "https://api.vaquill.ai/api/v1/us/statutes/section/{actId}" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"

curl "https://api.vaquill.ai/api/v1/us/statutes/section/{actId}/body" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"

Beyond search and retrieve, the same surface exposes citation resolve, section changes, cited-by, cross-state comparison, definitions, neighbors, and batch get. Cross-state is the one worth knowing about here, because "what does every state constitution say about this" is the question this corpus exists to answer.

Freshness: a ballot calendar, not a news feed

State constitutional amendments move on a schedule. A legislature refers a measure or citizens qualify an initiative, and the measure goes on a ballot. Many of those ballots are November elections, and some states, such as Texas, vote on amendments in odd-numbered years too. Between elections, the text barely moves.

That is why the registered cadence here is monthly. A registered cadence describes the schedule, and lastRetrievedAt on each board tells you when it was last pulled. On 12 October 2026 the board list showed 49 per-state boards for this corpus, plus one all-states board. We found no separate board for Puerto Rico, the Northern Mariana Islands or American Samoa, so check the section's own source and amendment fields for those three.

GET /boards is free and lists every watchable source. Each board carries a corpusType, a state, a label, the registered cadence, a lastRetrievedAt timestamp, publishesSourceChangedOn, and changeSignal. Compare lastRetrievedAt against cadence before treating a quiet watch as meaning nothing changed: lastRetrievedAt records when we last checked the publisher, not when the law last moved. To be told about changes, subscribe to a board with POST /watches. A watch receives the change events a refresh records and can pull the before and after text for each one. It does not make the board refresh any faster, and the board's registered cadence is what sets how soon a change can surface.

For the change side, yearFrom and yearTo filter on the last amendment year the publisher credits, so they track the law rather than our rebuild date. One caution that applies across the whole corpus: about a fifth of sections carry no amendment credit at all, because some publishers print none, and those sections drop out once you set either bound.

The versioning model, stated precisely

One citation holds one stored text, the one in force now. For an earlier date, pass asOf=2019-06-01 to the body endpoint. The earlier text is rebuilt from the before-side of the first change we observed after that date, and the asOf block reports which version you got (source), what it rests on (basisChangeId), how far back the evidence reaches (observedFrom), and isBounded. A true isBounded means we observed no change affecting your date, which is not the same as there having been none, because change capture began long after the corpus did. A section we know changed but cannot rebuild comes back unavailable.

Alongside that you have amendment history and a lastAmendedYear on each section, the yearFrom and yearTo currency filter, change events captured at each refresh, and per-section diffs on watched boards. Amendment History and Point-in-Time Law explains the distinction between a reconstruction and a published edition.

If your product needs to render the version of a constitutional provision in force on a specific past date, treat the asOf result as a lead and confirm it against the state's own published record before you rely on it.

Where this sits in the wider corpus

State constitutions are one slice of a US primary law corpus that holds more than 5.3 million distinct sections across 56 jurisdictions. The federal side includes the US Constitution at 74 sections, alongside the CFR, the Federal Register, the US Code, and the Statutes at Large. The state side adds statutory codes, administrative regulations, court rules, and insurance bulletins.

An open dataset of the state codes, the US Code and the state constitutions is published at github.com/Vaquill-AI/open-us-law, with its ingestion scripts. Its README lists the data under CC BY-NC 4.0 (free for non-commercial use with attribution, a licence is needed for commercial use) and the scripts under Apache-2.0.

FAQ

How many state constitutions are there?

Fifty, one per state. The corpus also carries constitutions for Puerto Rico, the Northern Mariana Islands and American Samoa, so corpusType=STATE_CONSTITUTION covers 53 jurisdictions. The District of Columbia, Guam and the US Virgin Islands are not included because none has a constitution of its own.

Why is the District of Columbia missing from the state constitutions API?

DC has no constitution. Its organic law is the District of Columbia Home Rule Act, a federal statute, so the closest equivalent lives in statutory law. DC appears in the state statutory code, state regulations and state court rules corpora.

Which state has the longest constitution?

By section count in this corpus on 12 October 2026, Arkansas leads with 513 sections, followed by Texas at 503 and Oklahoma at 478. Section count is a structural measure, not a word count, so it tracks how a state amends its constitution as much as how much subject matter it covers.

Are state constitutions more protective than the US Constitution?

They can be. A state constitution can grant more protection than the federal one but never less, because the federal document sets a floor. Several states have express privacy provisions with no direct federal analogue, which is why a search that stops at the US Constitution has checked the minimum only.

How often do state constitutions change?

Far more often than the federal one, because states generally amend through a legislative referral and a popular vote, and a number of states also allow a citizen initiative. Many amendments go to voters in November, and some states vote in odd-numbered years too. The registered refresh cadence is monthly, and GET /boards shows how each source is actually doing.

Can I retrieve a state constitution as it read on a past date?

In part. A citation returns the provision as it stands today by default, and asOf=YYYY-MM-DD on the body endpoint rebuilds an earlier text from the changes observed after your date. The asOf block reports isBounded when the answer rests on when capture began, and a date before that can come back unavailable. Amendment history, a lastAmendedYear per section, a yearFrom and yearTo currency filter, and change events per refresh round out the picture.

How do I filter a search to a specific article of a state constitution?

Pass article alongside corpusType=STATE_CONSTITUTION and state on POST /us/statutes/search, for example "article": "Article IV" with "state": "nm". Get the article labels from GET /us/statutes/divisions?corpusType=STATE_CONSTITUTION&state=XX and pass one back exactly as the listing returned it. state is required with article, because article labels repeat across constitutions and most states have an Article IV. A search hit's parent object already carries its article, so you can also spread a hit's parent into the next search to stay inside the same article.

Where does the text of each state constitution come from?

Results carry a source URL for the section, which is where to check the text. For most states it is a legislature or official code site, but American Samoa's is its bar association's page (asbar.org), so look at who runs the site before you treat it as the official text. Treat the API as the way to find and structure the text.

What does a state constitutions search cost?

A search call is 4 credits and a browse call is 1, per the pricing endpoint on 12 October 2026. The coverage endpoint and the boards endpoint are free. Failed work bills 0, and paging is cut from a single ranking, so results never repeat or go missing between pages.

Priyansh Khodiyar

Co-Founder & CTO

Priyansh leads engineering and AI at Vaquill AI: the pipelines that pull statutes, regulations and court rules from every US jurisdiction's official publisher, and the REST API, MCP server and open dataset that serve them.