DMV Knowledge Test Pass Rates by State: What Is Actually Published

Every few months a list goes around ranking the "hardest DMV tests in America," and every version of it has a number for all fifty states. Most states do not publish a knowledge-test pass rate. The lists are filling in the blanks.

This page does the opposite. It shows the pass rate for every jurisdiction that has one on the record, says exactly where each number came from and what it measures, and leaves the rest blank. Fourteen states have no published figure that could be traced to their licensing agency. Those cells say so. If you came here to find out whether your state's test is hard, the blank cells are part of the answer, and the section after the table explains why the numbers that do exist tell you less about your own attempt than you would expect.

How these numbers were collected

Three kinds of source were accepted, in this order of strength:

1. The agency itself. A licensing agency's published statistics page, annual report, news release, or a statement on its own website. Oregon's DMV publishes knowledge-test attempts and pass rates by year. Minnesota's Department of Public Safety stated a failure rate in a 2024 news release. Maryland's Motor Vehicle Administration states one on its practice-test page. Arkansas State Police published the 2019 counts in a 2021 release.
2. A records analysis that names the agency as the source. In April 2024, USA TODAY's Blueprint team requested 2020-2023 test results from every state's licensing agency and published the results for the 36 states that provided them, including a per-state knowledge-test pass rate. Fourteen states did not respond or declined. That analysis is the single largest source in the table, and it is treated as what it is: a pooled four-year figure covering every test administered, with first attempts and retakes mixed together.
3. A news report that attributes a specific figure to the agency. A D.C. outlet citing "according to the DMV," a New Jersey Motor Vehicle Commission official quoted in the commission's own release, a Tulsa World analysis of Department of Public Safety testing records.

What was rejected: every pass rate that appears only on a test-prep site, a driving-school blog, or an aggregator without a traceable agency source. Several of those sites list a precise-looking figure for all fifty states. When a figure could not be traced past the site that published it, it was not used, even where it looked plausible. Where a state's agency was searched and nothing was found, the table says "Not published" rather than borrowing a number from somewhere else.

Three caveats apply to every row, and they matter more than the decimals:

• First attempt versus all attempts. A state that reports "55% of tests passed" is counting retakes. A state that reports "half of first-time takers fail" is counting people. The two are not comparable, and a state whose applicants retake quickly and often will show a lower all-attempts rate for the same test. The "what the number covers" column says which is which. Where the source did not say, the column says so.
• Year. Several figures are pooled across 2020-2023. Oregon's own series shows how much a rate can move inside a few years. A pooled figure is an average of a moving target.
• Test version. Some states changed their test length or moved to online testing inside the measurement window. A pass rate from the 25-question era of a test does not describe the 40-question version.

Pass rates by state

Rates are shown exactly as published or as computed from published counts (computed figures say so). A state with no row is a state where no traceable figure exists, not a state with a secret. The source link on each row goes to the document the figure was checked against.

What the table actually shows

Read the published rows together and a few things stand out.

The spread is wide, and it is not about the rules. Among the states in the USA TODAY records analysis, the pooled knowledge-test pass rate runs from the high 30s to the high 80s. Vermont, with a 20-question test that costs $39 every time you sit it, is at the top. Missouri, Wyoming, Indiana and California are at the bottom. The traffic laws in those states are not wildly different. What differs is test length, how the questions are written, whether applicants are tested after a required course, how cheap and fast a retake is, and who is in the pool. A state where a retake is free and same-day will see more people walk in cold and more attempts counted.

Short tests do not mean easy tests. Pennsylvania's 18-question test has one of the higher published pass rates in the table. Vermont's 20-question test has the highest. But the margin on a short test is small: miss four on Pennsylvania's test and you fail. People who pass short tests at high rates are, mostly, people who studied because the small margin was obvious. How to work inside that margin is the subject of How to pass the Pennsylvania knowledge test on your first try.

The rate can fall fast. Oregon's DMV publishes its own series. In 2021, 54% of knowledge-test attempts passed. In 2022 it was 45%, in 2023 41%, in 2024 40%, and in 2025 42%. Same state, same agency, a drop of a dozen points in three years. The pooled 2020-2023 figure reported to USA TODAY for Oregon was 57%. Both numbers are real; they measure different windows. If you are reading a list that gives one static number per state, ask which year.

The low-rate states are the ones that count attempts. Several of the lowest published rates are explicitly all-attempt figures. Oregon's DMV says plainly that its rates "do not reflect how many individual customers took the test, or how many attempts an individual customer made." A person who fails three times and passes on the fourth contributes one pass and three fails. That pulls the rate down without telling you anything about how a prepared first-time taker does.

Why Washington's test is famous for being hard, and what the number says

Ask around and Washington's knowledge test has a reputation as the hardest in the country. The reputation has a source. In 2012, the Department of Licensing told a Seattle reporter that about half of the people who took the written test failed on the first attempt, and that figure has been repeated, rounded, and reattached to later years ever since.

There are real reasons the Washington test feels different, and most of them are about who runs it. The state does not give the knowledge test at a licensing office. It is administered by approved driver-training schools and testing locations, each of which sets its own fee, so the first thing many applicants learn is that the test costs whatever the nearest school charges. The test is 40 questions, 32 correct to pass, and the Washington Driver Guide prints neither of those numbers; you have to find them on the Department of Licensing's testing page. Uncertainty about the format is a large part of why a test feels hard.

The retake rule is the clearest example. The Department of Licensing says only that there "may be a waiting period" before you can retest, and never says how long. That is not vagueness for its own sake: state rule requires each approved school to set and disclose its own retesting policy, so the waiting period is whatever your testing location's policy says. There is no statewide number to publish. Ask the school before you leave, because the answer is theirs, not the state's. One statewide limit does sit behind it: an applicant disqualified on the knowledge test three or more times loses eligibility for a driving-test waiver.

The published number tells a quieter story. In the 2020-2023 records analysis, Washington's pooled knowledge-test pass rate was about 58%: roughly the middle of the table, above California, Oregon, Indiana, Missouri and Wyoming, below Pennsylvania, Vermont, Louisiana and Kansas. Washington's test is not easy. It is also not, by the only recent number that can be traced to agency records, the hardest. The "hardest in America" label is a 2012 first-attempt statistic that has outlived its year.

If you are testing in Washington, the practical lesson is the one that applies everywhere: find out the format before you sit down (40 questions, 32 to pass, fee set by the testing location, ask about the wait), and study the state's own rules, not a national question bank. The Washington-specific rules that trip people are the same kind that trip people in every state, and the answer to "is this test hard" is mostly "did you study the right things."

What a pass rate does and does not mean for you

A state pass rate is the wrong instrument for the question most people bring to it. It does not predict your result. Three things do.

Whether you know the rules as the test asks them. Knowledge tests do not ask "is speeding bad." They ask which of four near-identical statements about a four-way stop is correct, or what the number is for following distance behind an emergency vehicle. A rule you half-know is a rule you will miss when the choices are written to be close. The right-of-way topic page is the best example: the rule is simple, and the question patterns are written to make you second-guess it.

Whether you know your own state's numbers. Blood-alcohol limits for drivers under 21, new-driver passenger restrictions, phone laws, insurance minimums, the fee and the wait if you fail. These change state to state and they are the questions a generic national practice app gets wrong. The state pages (Oregon, California, Pennsylvania, Florida) carry them with the handbook citation beside each fact, and the full topic list is at /topics.

Whether you read the question. The cheapest way to fail is to answer a question the test did not ask: miss the word "except," answer for a flashing red when the question said flashing yellow, stop reading the options at the first one that sounds right. No pass rate captures this, and it is the most fixable failure there is. The wrong answers are usually built that way on purpose, which the ten ways a permit-test question is built to catch you walks through one trap at a time.

One more thing a pass rate cannot tell you is how you will hold up on the day. If the number you are really looking for is reassurance, keeping nerves from costing you points is the more useful read, because it works from the parts of test day that are actually published.

What a pass rate is good for is setting your expectations about the retake. Where a state's published rate is in the 40s or 50s and counts all attempts, a large share of the people in line with you have failed at least once, the office handles it routinely, and the retake rules are built for it. The waiting periods and fees for every state are in the retake rules comparison, and the plan for the days between a fail and a retake is in Failed the DMV test? What to do next.

Why so many states do not publish this

The blank cells are not an accident of searching. Chasing each one to its source shows three different reasons a state has no number, and they are worth separating.

Some states count the other test. Texas publishes a monthly driver-license data set on the Department of Public Safety site that reports tests passed and failed, which sounds like exactly the right document. Every month of it reports skills tests: non-commercial road tests and the commercial vehicle-inspection, basic-control and road tests. There is no knowledge-test line in any issue. Hawaii is the same story from the other direction: Honolulu's customer-services department has given reporters road-test pass rates for years, and no county publishes a written-test rate.

Some states count how many tests, not how many passed. Tennessee's annual reports have published knowledge-test volume for over a decade (231,218 Class D and M knowledge tests in one year; more than 117,000 teenagers taking it online in another) with no pass-fail split anywhere. Wisconsin's 2025 announcement of at-home testing says the DMV administers over 4,000 Class D knowledge tests a month, and stops there. West Virginia's releases count online tests taken, not passed.

And some states genuinely do not look. This is the most revealing category. A Tennessee comptroller's audit of the Department of Safety recorded that management does not analyze exam results to judge whether a question was too hard or unfair. Michigan's Auditor General published a performance audit of branch operations in 2025 with a finding devoted entirely to how offices administer the knowledge test; it examined proctoring discipline and reported no pass or fail statistics at all. Alaska's DMV runs a real statistics program, with a licensed-driver series going back to 1998, that has never included a test outcome.

That last group matters for how you read the whole table. A pass rate is not a byproduct that every agency automatically has sitting in a drawer. Where a number exists, some agency chose to count and publish. Where one does not, the absence usually means nobody is measuring, not that the number is embarrassing.

It also explains why so many websites can quote a figure for all fifty states. Precise-looking rates circulate for Alaska, Massachusetts, New Mexico and Wisconsin among others. Each one traces back only to a test-prep or driving-school page, never to an agency. There is no hidden federal table these sites are reading; for fourteen states, the number does not exist in public.

A note on the blanks

Fourteen states have no traceable knowledge-test pass rate in this table: Alaska, Connecticut, Hawaii, Massachusetts, Michigan, Mississippi, New Mexico, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, West Virginia and Wisconsin. Most of them did not provide data to the 2024 records request (Texas provided road-test results only), and no agency report, legislative audit, or agency-attributed news figure could be located for any of them. Several of those states have a precise-looking pass rate circulating on test-prep sites. None of those figures could be traced to the agency, so the blank stands. If your state is blank and a website is telling you its pass rate to one decimal place, ask where the number came from.

Where an agency's own figure and the 2024 records analysis both exist, the agency's own figure is the one in the table and the other is noted beside it. Minnesota is the clearest case: the Department of Public Safety said in 2024 that 56 percent of people who take the written test fail, while the 2020-2023 pooled records put the pass rate near 60 percent. Both are agency-sourced. They measure different windows and, as far as the agency has said, possibly different populations. That gap is the whole argument for reading the caveat column before the number.

This table is refreshed when a state publishes something new. Oregon publishes annually; most states publish nothing. The "last updated" date at the top is the date the sources were last checked.