AP Score Distributions: What Percentage Gets Each Score

AP score distributions are the only official statistic College Board publishes about how a full cohort performed: the exact percentage of students who earned each score from 5 down to 1. This page puts the 2026 AP score distribution for all 40 exams in one table, links every exam we model to its calculator, and then explains the part most rankings skip — why comparing raw 5 rates across exams tells you far less than it appears to.

What AP Score Distributions Actually Show

An AP score distribution is a headcount expressed as percentages: the share of test-takers who earned each score from 5 down to 1, published once every exam in that administration has been scored and equated. It is not a curve, not a raw-score conversion, and not a difficulty rating.

That distinction matters more than it sounds. The composite cutoffs on our AP exam score calculators — the share of raw points you need for a 5 — are estimates, reconstructed from released exams and reported results. AP score distributions are the opposite: College Board publishes them itself, they are exact, and nobody has to reverse-engineer them. When the two seem to disagree about which exam is “easier,” the distribution is the fact and the cutoff is the inference.

Below is the 2026 AP score data for all 40 exams, sorted by the percentage earning a 5. Everything after the table is about what those numbers can and cannot support.

When AP Score Distributions Are Released

Scores reach students in early-to-mid July. The distributions follow within days, in two waves. Packer posts figures exam by exam through the first half of July, and the same numbers appear on College Board's per-exam pages at or near the same time. By late July the full roster is normally out.

The 2026 numbers below are those first-wave figures. They are published as whole percentages and shown in italics on College Board's site to mark them preliminary — compare them with the 2025 rows on the same pages, which carry a decimal place. The detailed per-exam PDFs that add test-taker counts, mean scores, and one-decimal precision arrive later.

One practical consequence: 2026 test-taker counts are not yet posted for most exams. Where Packer stated a 2026 count publicly we use it and label it 2026. Otherwise the table shows the 2025 final count, clearly labeled, or says the number has not been posted. We would rather leave a cell honest than fill it.

How to Read an AP Score Distribution

Start with the number everyone quotes: the share earning a 5. Across the 40 exams it runs from 48% (AP Chinese Language and Culture) down to 7% (AP 3-D Art and Design). Within the 21 exams we build calculators for, the spread is 46% (AP Calculus BC) down to 10% (AP Computer Science Principles).

The instinctive reading — that Calculus BC is four and a half times easier — is wrong, and it is wrong for a reason worth understanding. A distribution is two things multiplied together: how demanding the exam is, and who sat it. From the outside those two inputs cannot be separated, and on several exams the second one dominates completely.

AP Calculus BC is the clearest case. Its cohort has already finished a year of calculus, self-selected into the harder of the two calculus courses, and is largely bound for quantitative majors. A 46% 5 rate on that population is not evidence of a gentle exam; it is evidence of a filtered, well-prepared group. The same logic covers AP Physics C: Mechanics and AP Physics C: E&M, which post 72% and 75% pass rates on cohorts a fraction the size of AP Physics 1.

AP Computer Science Principles runs the other way. It is designed as a first, accessible computing course and is offered widely to students with no programming background. A 10% 5 rate on a broad, open-enrollment cohort is a different measurement than the same figure on a filtered one. It also happens to carry the toughest composite cutoff we track — roughly 83% of available points for a 5 — so there the exam and the cohort push in the same direction.

The working rule: compare an exam against its own past years, or against exams with a similar enrollment profile. Comparing AP Calculus BC to AP Computer Science Principles tells you almost nothing about which one you will find harder.

Percentage of 5s by AP Exam: Three Traps

Heritage speakers. The two highest 5 rates on the board — AP Chinese Language and Culture at 48% and AP Japanese Language and Culture at 47% — are Total Group figures, which include students who speak the language at home. Roughly 70% of AP Chinese test-takers are heritage speakers. Packer has noted that about 72% of AP Spanish Language students and about 25% of AP French students have meaningful exposure beyond the classroom. College Board publishes a separate standard group distribution for precisely this reason. Reading a total-group 5 rate as proof an exam is easy is the most common misuse of AP score data there is.

Exams with no composite at all. Five of the 40 — AP Seminar, AP Research, AP 2-D Art and Design, AP 3-D Art and Design, and AP Drawing — are portfolio and performance-task assessments. There is no multiple-choice section, no free-response raw total, and therefore no “percent of the composite needed for a 5.” AP Research posts the highest pass rate of any AP exam at 90% scoring 3 or higher and AP Seminar is close behind at 88%, but those numbers describe a submitted body of work judged against a rubric, not a three-hour test. That is also the honest reason our curve-forgiveness comparison stops at 21 exams while this distribution table covers all 40.

Small cohorts. AP Italian had 2,241 test-takers in 2025, AP Japanese 3,245, AP German 4,213, AP Latin 4,336. On numbers that small a few hundred students move a percentage several points, so year-over-year swings that look dramatic frequently are not. AP Latin's 3+ rate jumped from 58.6% in 2025 to 73% in 2026 and its 5 rate from 12.5% to 20% — the largest single-year move on the 2026 board. It is real and it is straight off College Board's page. It is not a reason to build a schedule around AP Latin.

The AP Computer Science A Bimodal Distribution

Most AP score distributions are roughly hill-shaped: a modest tail of 1s, a bulge across 2 and 3, a thinner 4, a smaller 5. AP Computer Science A is not. In 2025 it produced 25.6% 5s and 22.0% 1s — the two ends of the scale were almost the same size, with the middle scores squeezed between them. The 2026 preliminary figures show the same shape.

This is the signature of a skills-based programming exam. The free-response questions ask you to write working Java: a method, a class, a 2-D array traversal. Code either implements the required logic or it does not, and partial credit rewards correct structure rather than a good-faith gesture. A student fluent with arrays, ArrayLists, inheritance, and recursion can attempt all four questions; a student who never got comfortable with the syntax often cannot start them. There is far less middle ground than on an essay-based exam, where a partial argument still earns partial points.

The practical implication is that the headline 5 rate is not an invitation. The same exam that produces a quarter 5s produces nearly a quarter 1s, and its overall pass rate (66%) sits below AP Biology, AP Chemistry, and AP U.S. History. AP Computer Science Principles, which shares a subject but not a format, has a far more conventional shape: 10% 5s and a 63% pass rate.

Distribution vs Curve: Two Numbers People Confuse

A score distribution and a scoring curve are different objects, and conflating them produces most of the bad advice online. The distribution says what fraction of students landed on each score. The curve — properly, the composite cutoff — says how many raw points you needed to get there. An exam can have a forgiving cutoff and a mediocre distribution, or the reverse.

AP Statistics is the cleanest example. Its estimated cutoff for a 5 sits near 71% of the composite, squarely mid-pack among the 21 exams we model. Its 2026 pass rate, 62%, is the second-lowest of all 40 exams, behind only AP Music Theory at 59%. Nothing about the cutoff explains that. The cohort does: AP Statistics draws a broad enrollment that includes many students taking it as a terminal math course.

Running the other way, AP Calculus BC's cutoff for a 5 is roughly 57% of the composite — the most forgiving we track, and 26 percentage points below AP Computer Science Principles at about 83% — and its distribution is strong too. When cutoff and distribution agree, you are usually watching cohort strength show up twice.

How to Place Yourself in the Data

1. Score a released practice exam honestly against the official rubric — no partial credit a reader would not actually award you.
2. Convert your raw points into a composite using the matching tool on our AP exam score calculators page.
3. Read the estimated 1–5 result, then treat anything within about three composite points of a cutoff as a coin flip rather than a prediction.
4. Find your exam's row in the table above to see what share of last year's cohort landed where you did.
5. Repeat with a second released exam before you trust the answer. One practice test is a data point, not a trend.

How We Normalized These Numbers

Two versions of the same statistic circulate every summer, because College Board's pages carry both the new preliminary year and several historical years, and aggregators mix them freely. Our own calculator pages were built during the 2025 cycle, and several carried 2025 figures with a decimal place — AP Biology at 18.9% 5s, AP Chemistry at 17.9%, AP Psychology at 14.4%.

For this table we normalized every row to 2026 so that a 40-exam ranking is internally consistent. The values that moved: AP Biology to 15% and 71% 3+, AP Chemistry to 15% and 75%, AP Environmental Science to 13% and 69%, AP Psychology to 15% and 74%, AP World History to 14% and 66%, AP U.S. History to 14% and 74%, AP Computer Science A to 25% and 66%, and AP Statistics to 17% and 62%. Rows that were already whole numbers were already 2026 and did not change.

Two standing caveats apply to everything above. The distributions are official published figures. The composite cutoffs quoted alongside them are not: College Board has never released an official raw-score-to-1-5 conversion for a live exam, and it re-equates every exam every year, so every cutoff on this site is an estimate that moves.

Where to Go Next

If you have a raw practice score and want the 1–5 it implies, pick your exam from the full set of AP exam score calculators — 21 of the 40 exams above, each with its own researched cutoffs and score distribution. The heavily used ones are AP U.S. History, AP English Language, AP Calculus AB, and AP Psychology.

For the reasoning behind the cutoffs rather than the results, the rest of our guides cover how the curve is set, which exams are most forgiving, and why “forgiving” and “easy” are not the same word.

Frequently Asked Questions