Six numbers are published. The one everybody quotes is not among them

The number everybody quotes does not exist.

Search for the CPA exam pass rate and the answer comes back inside a second: somewhere around 50 to 55 percent, with a table of section rates underneath it. Google's own summary panel carries that band. So do the review-course pages ranking beneath it, the university blogs, and the state societies. The band is repeated so consistently that it reads as a published statistic.

The AICPA page that everyone is citing publishes six numbers and no overall figure at all.

What is the CPA exam pass rate in 2026?

Answer: The AICPA publishes a rate for each of the six sections. Through the first half of 2026 the cumulative rates are 48.65 percent for Auditing and Attestation, 42.95 for Financial Accounting and Reporting, 66.78 for Regulation, 43.64 for Business Analysis and Reporting, 67.45 for Information Systems and Controls, and 79.53 for Tax Compliance and Planning.

The equivalent figures for full-year 2025 are 48.21, 42.12, 63.12, 41.94, 67.79 and 77.65. Every candidate sits the three Core sections, which are AUD, FAR and REG, and chooses one of the three Disciplines. A passing score is 75 on a scale that runs from 0 to 99, and the AICPA states plainly on the same page that the reported score is not a percentage correct and that scores are not curved.

The spread across those six numbers is 36.58 points. Tax Compliance and Planning passes about four candidates in five. Financial Accounting and Reporting passes about two in five. Whatever a single overall rate is supposed to summarise, it is summarising a set of numbers that do not sit near each other.

Why is there no official overall CPA exam pass rate?

Answer: Because there is no obvious way to compute one. Each candidate takes all three Core sections and only one of three Disciplines, so an average across the six treats a section every candidate sits as equal in weight to a section a minority sits. The AICPA declines to publish an aggregate, and the figure in circulation is somebody else's arithmetic.

Take the unweighted mean of the three Core sections for the first half of 2026 and the answer is 52.79 percent. Take the unweighted mean of all six and it is 58.17 percent. For 2025 the same two calculations give 51.15 and 56.80 percent.

Only one of those lands in the band that gets quoted. The 50 to 55 percent figure is the Core-only average, which means the number presented as the overall pass rate of the CPA exam silently discards half of its sections, including the highest-passing one.

What would a properly weighted pass rate look like?

Answer: Weight each section by how many attempts it generates. A section with a 43 percent pass rate produces roughly 2.33 attempts for every pass, so it contributes more attempts to the annual pool than a section that passes 80 percent of the time. On the first-half 2026 rates that puts the attempt-weighted rate at 52.9 percent.

The calculation is short enough to check. If a section passes a fraction p of attempts, the expected number of attempts a candidate needs is 1 divided by p. For the first half of 2026 that is 2.06 attempts at AUD, 2.33 at FAR, 1.50 at REG, 2.29 at BAR, 1.48 at ISC and 1.26 at TCP. A candidate who eventually qualifies makes four passes. Add the expected attempts for the three Core sections and one Discipline, assume the three Disciplines are chosen in equal thirds, and the total is 7.56 attempts for 4 passes. That is a rate of 52.9 percent. Run the same arithmetic on the 2025 rates and it gives 51.6 percent.

Here is where the number stops being a fact about the exam. Hold the section rates fixed and change nothing except which Discipline candidates choose. If every candidate took Business Analysis and Reporting the equilibrium rate would be 48.9 percent. If every candidate took Tax Compliance and Planning it would be 56.0 percent. The whole range that gets quoted as the difficulty of the CPA exam can be traversed by candidate preference alone, with the exam untouched.

Two assumptions in that model deserve to be stated where they matter rather than parked at the bottom. It assumes every candidate eventually passes, which is false, and it assumes a constant per-attempt probability, which is also false, because re-takers and first-time candidates do not perform alike. Both push the real figure below the model. What the model establishes is not the true rate but the shape of the thing: an aggregate pass rate on a four-section exam with elective content is a composite of candidate behaviour, and it was never a measurement of difficulty.

Does a rising pass rate mean the exam got easier?

Answer: The AICPA says no, and says so in a sentence worth reading twice. Its pass-rates page states that the exam "is not harder or easier to pass at different times" and that "an increase in pass rates simply means that candidates are better prepared." Scores are equated across test forms, so within a version of the exam the rate is a property of who sat, not of what was asked.

Accept that claim and an awkward consequence follows immediately. A statistic that moves only with candidate preparedness is a statistic about the candidate population. It is a measure of the pipeline, wearing the clothes of a measure of the test.

Across versions the picture changes again, and the AICPA has been explicit about it. Its February 2025 article Trust in the CPA Exam records that the passing scores now in force were set by the Board of Examiners in spring 2024, that the sections "are not designed to have the same pass rate as the previous CPA Exam", and that drawing comparisons to previous years "provides limited information". The same article says the full effect of CPA Evolution on volumes and pass rates "will not be known until late 2026".

So the rate is equated within a version and reset between versions. Neither property is visible in a table of six percentages, and the pages that stack 2019 through 2026 into one chart are joining series the publisher has said should not be joined.

What did the pass rate fail to show about 2025?

Answer: The largest change in the CPA pipeline in a decade. NASBA's count of candidates passing their final section rose to 23,978 in 2025 from 13,070 in 2024, a rise of 83.5 percent. Section pass rates over the same stretch moved by an average of 1.48 points. The rate registered almost none of it.

The rest of NASBA's arithmetic sits in the same direction. Its 2025 report, announced on 18 August 2026, counts 87,660 candidates taking the examination against 74,165 in 2024, a rise of 18.2 percent, and 36,590 new candidates against 27,994, a rise of 30.7 percent. Divide the finishers by the candidates and the share who completed the exam went from 17.62 percent to 27.35 percent in one year.

The mechanism is in the AICPA's own list of transition policies, and none of it is about difficulty. Credits held on 1 January 2024 were extended by a one-time policy to 30 June 2025, which put a hard completion deadline in the middle of the year. State boards moved candidates to credit windows of 30 months or more. A Credit Relief Initiative restored credits that had expired between 1 January 2020 and 13 May 2023 during the pandemic emergency. Core sections returned to continuous testing in 2025.

A cohort was pushed through a gate. The pass rate cannot see a gate.

Who is actually sitting for the exam?

Answer: Fewer people than a decade ago, drawn from a shrinking pool of accounting graduates, and the composition of that pool moves the rate on its own. NASBA's calendar-2024 data show pass rates falling with age, from 64.1 percent for candidates under 22 to 43.5 percent for candidates aged 30 and over.

That spread is 20.6 points, which is larger than the gap between AUD and REG. It is the clearest available evidence that the rate tracks who sits. A year in which more career-changers and more re-entrants take the exam will print a lower rate than a year dominated by candidates straight out of a master's programme, with the exam identical in both.

Upstream, the AICPA's 2025 Trends release counts 55,152 accounting bachelor's and master's graduates in the 2023-24 academic year, down 6.6 percent, following declines of 9.6 and 7.4 percent in the two prior years. Provisional federal data put the bachelor's figure at 40,817, down 3.3 percent, while master's degrees in accounting or taxation fell about 15 percent to 14,335. The same release reports two consecutive semesters of 12 percent growth in accounting enrolment in 2024-25, which is a leading indicator pointing the other way.

New candidate entries tell the transition story cleanly: 42,626 in 2023, the highest since 2016 as candidates rushed to finish under the old exam, then 28,082 in 2024. Global sections taken fell from roughly 204,000 to 148,000 across the same two years, a drop of 27.5 percent. Two sections per candidate, near enough, in the year the model changed.

One small thing is worth noticing about those counts. The AICPA puts new 2024 candidates at 28,082 and NASBA puts them at 27,994, a gap of 88 people in a figure the two bodies produce from the same examination. Neither publishes the definitional basis that would reconcile them. The pipeline numbers are administrative counts with edges, and they should be quoted as such.

Why do review courses publish pass rates near 90 percent?

Answer: Because they are measuring a different population and, to their credit, most of them say so on the page. The published national rate counts every attempt at a section. A provider rate typically counts students who completed a defined amount of the course and reported their result.

UWorld's page states that "students who meet all SmartPath targets have a 90% pass rate" and that the figure "is based on self-reported CPA Exam results from students using our SmartPath Predictive Technology". Becker's page defines its Exam Day Ready cohort as students who watched 80 percent of the lecture videos, answered 80 percent of the questions and simulations correctly, cleared 50 percent on five practice exams, then "sat for an exam and reported their score". Surgent's page carries a "92% CPA Exam pass rate" beside a note that national rates hover around 45 to 55 percent.

Read the definitions rather than the percentages and the difference is not a dispute about facts. A student who watches four fifths of the videos, answers four fifths of the questions correctly and then chooses to report a score is a student the course has already screened twice. That is a legitimate thing to measure. It is not the same measurement as the national one, and the two should never be set side by side as though the gap between them were an effect.

The pages themselves are not fully consistent on each other. Becker's page attributes an 88 percent claim to Surgent; Surgent's own page said 92 percent when both were checked on 22 August 2026. Becker's page also attributes a 91 percent claim to Gleim, and no such figure appears on Gleim's pass-rates page.

Do the published pass rate tables even agree?

Answer: Not exactly. Gleim's page, which ranks on the first page of results for this query, prints a 2025 cumulative column that differs from the AICPA's on five of six sections. The gap is largest at ISC, 66.71 against 67.79.

The reason is reproducible. Average the four quarterly rates that the same table prints and you get 48.0400, 42.1150, 63.0975, 41.0175, 66.7125 and 77.2425, which match all six of its cumulative figures to two decimals. That column is the unweighted mean of the quarters. The AICPA's cumulative is the rate for the year, which weights each quarter by the attempts made in it.

The direction of the gap is itself informative. The AICPA's figure sits above the simple mean on all six sections, so the quarters carrying the most attempts were also the better-performing quarters. The two Discipline sections show the largest divergence, at 1.08 and 0.92 points, which is what happens when quarterly volumes are uneven. The same page gives Tax Compliance and Planning at 79.95 percent for the second quarter of 2026 where the AICPA gives 79.75; the AICPA's own quarterly pair reconciles with its printed cumulative and the other pair does not.

None of this is an accusation about anybody. It is a note about what happens to a statistic as it is copied: a table that looks official three links deep is a table somebody re-derived, and the re-derivation is checkable in a spreadsheet.

There is a reason those pages exist. The AICPA's own page carries 2025 and the first two quarters of 2026 and nothing before that. The longer history of this statistic, the part that would let anyone judge whether 42.95 percent at FAR is unusual, is held by the companies selling preparation courses.

What the number is good for

The section rates are a real measurement and they answer a real question, which is how a given attempt at a given section is likely to go. That is useful to a candidate sequencing four exams around a job, and it is the question the AICPA set out to answer.

Every other use of the figure is a substitution. It gets quoted as the difficulty of the exam, which it is not, because the exam is equated and the rate moves with who sits. It gets quoted as the health of the profession, which it cannot be, because it moved 1.48 points in a year when the number of people finishing the exam rose 83.5 percent. It gets quoted as a single national percentage, which no institution publishes.

The numbers that carry the profession are the ones underneath: 87,660 sitting, 36,590 arriving, 23,978 finishing, 55,152 graduating and falling. Those are counts of people. Set them next to a rate that counts attempts and it becomes clear which of them was ever going to tell you something.

Frequently asked questions

Is there an official overall CPA exam pass rate? No. The AICPA publishes a cumulative rate for each of the six sections and no aggregate. Any single overall figure is a computation somebody performed on those six, and the common 50 to 55 percent band is the unweighted average of the three Core sections only.

Does the pass rate tell me my odds of becoming a CPA? No. It is a rate on section attempts, so a candidate who fails a section twice before passing contributes three attempts to it. No published figure states what share of candidates eventually pass all four sections.

Which CPA exam section has the lowest pass rate? Financial Accounting and Reporting, at 42.95 percent cumulative through the first half of 2026, narrowly below Business Analysis and Reporting at 43.64 percent. Tax Compliance and Planning is highest at 79.53 percent.

Can pass rates from before 2024 be compared with today's? The AICPA advises against it. Its February 2025 article states that the sections under CPA Evolution are not designed to carry the same pass rates as the previous exam and that comparison with prior years provides limited information. Its own published table retains only 2025 onward.

Why did so many more candidates finish the exam in 2025? NASBA counts 23,978 candidates passing their final section in 2025 against 13,070 in 2024. The AICPA's transition policies put a completion deadline of 30 June 2025 on credits held at the start of 2024, extended credit windows to 30 months or more, and restored credits that expired during the pandemic emergency. Section pass rates moved an average of 1.48 points over the same period.