Methodology

The one rule everything else follows from

A figure earns a place on this site when someone has read it on a named source. Not when it is widely repeated, not when three test prep sites agree, and not when it sounds right. If nobody has read it on the licensing agency's own page, its official manual, the state vehicle code, or the state administrative rules, it does not render.

That rule sounds obvious and it is expensive. It is the reason 17 states on this site do not carry a confirmed question count: their agencies genuinely do not publish one. Several of those were checked more than once, including full text searches of the current driver manual, before being left blank. South Dakota's manual runs to seventy two pages and does not contain the word question. Wyoming's runs to ninety eight and does not either.

What a source has to be

In descending order of what we will accept:

• The statute or the administrative rule. A pass mark written into a state code is the strongest thing available, because it cannot change without a public process. Hawaii's eighty percent is in the administrative rules. Mississippi's is in title 31 of its code.
• The licensing agency's own page or its official manual. Second best, and the most common source here. Agency pages change without notice, which is why the date matters.
• Another named government body with first hand knowledge. Where county agencies run licensing, as in Hawaii, the county page is the agency page.
• Nothing else. Test prep sites, driving school blogs, news write ups and answer aggregators are not sources of record here, however confidently they state a number. Where our figure traces only to one of those, the figure does not appear and a note says why.

The three things every figure carries

Every sourced fact in the data files carries a source name, a source URL and an ISO date recording when it was last read against that source. Continuous integration refuses the build if any of them are missing, so it is not possible to add a number to this site without saying where it came from. The same check enforces that a state which is not confirmed carries a written note explaining exactly what could not be confirmed.

Federal citations are checked by machine

The commercial questions cite federal regulation, and a confident citation to a section that does not exist is worse than no citation at all. So the build pulls the current structure of title 49 from the eCFR and checks that every federal section the banks cite is really there, printing each one beside its official heading.

It paid for itself immediately. It caught a question citing 49 CFR 392.31 for a drawbridge stopping rule: that section does not exist, and Part 392 carries no drawbridge provision at all, so the question was deleted rather than patched. A second pass reads the regulation text itself and compares it against what our answers claim, which is how two air brake questions were caught asserting a low air pressure warning at roughly sixty psi when 49 CFR 393.51 sets it at fifty five psi or half the governor cutout pressure, whichever is less. Sixty is the figure from a different standard.

What the tiers mean

Indexability here is an allowlist rather than a default. A page is in Tier A only when it has been deliberately added, and Tier A is the only thing in the sitemap.

• Tier A. Pages whose numbers are confirmed and whose content is substantial enough to be worth a stranger's time. State hubs, practice test hubs, written and road test pages for confirmed states, the national comparison tables, the sign reference, the guides, and this page.
• Tier B. Everything else, rendered with a no index directive and still fully usable. Individual test runner pages sit here because they are tools rather than documents. So does every state whose format could not be confirmed, however good its practice tests are.

A state does not lose its practice tests for being unconfirmed. It loses its place in search until an agency publishes something we can read. That is the honest trade: a learner in an unverified state still gets a full question bank, and Google does not get told we know something we do not.

What we refuse to publish

Office level pass rates

Pass rates by licensing office are the single most requested figure in this category and the easiest thing in the world to make up. Numbers circulate for individual test centres with no source at all, and they move people to drive across a state for a centre that may be no easier. We will publish this only from the agency's own records, obtained through a public records request in our own name. That request is drafted and waiting to go in. Until real records come back, this site publishes nothing at all on the subject, estimates included.

Numbers we could not confirm

Where a question count or pass mark is not published, the state page says so in plain language and names what is missing. It does not quietly show the figure that test prep sites repeat. Several states here have a widely quoted number that traces to nobody, and the difference between reporting that and repeating it is the whole point of the site.

Invented social proof

There are no user counts on this site, no testimonials, no star ratings, no pass rate claims about our own learners, and no badges asserting that anybody trusts us. Every one of those is trivially fabricated and impossible for a reader to check, and the moment a site prints one it has taught you to stop reading carefully.

Handbook text and real exam questions

State driver handbooks are copyrighted, and real exam items are secure test material. Neither is copied or closely paraphrased here. Questions are written from the underlying public law, which is why every one of them carries a statute or an official source rather than a page number.

A recommendation dressed as a fact

Pages that touch legal status, medical fitness or anything else where an individual needs professional advice record what the law and the agency say and stop there. The page on credentials issued without proof of lawful status is the clearest example: citations and agency wording, no guidance of any kind.

How the dates work

Verification is a repeated pass, not a one off. A figure keeps its original date until somebody reads it against the source again, at which point the date moves and the change appears in the log. Because the log is generated from the data files rather than written by hand, it cannot drift away from what the site actually claims.

The current record spans 2 verification dates, from 2026-08-27 to 2026-08-28. Read the log to see what moved and when.

When an agency site will not let us in

Some state domains block automated reads outright. When that happens the row says so rather than pretending otherwise, and it names the alternative that carried the fact, usually the statute or an archived copy of the official manual. Illinois, Michigan and Montana have all blocked reads recently, and the notes on those rows say which pages were refused on which date.

Corrections

If a figure here is wrong, the fastest fix is a link to the agency page that proves it. Corrections are made against the source, the date moves, and the change shows up in the log the same day. There is no version of this site where a correction is quietly applied without the date changing.

Who writes this site, and the full verification log.