What Your AFQT Score Really Is: From Four ASVAB Subtests to a Percentile

Recruiters talk about the AFQT as if it were a test score: you need a 31, a 50 opens more doors. It is not a score in that sense. It is the end of a short chain of published arithmetic that starts with four ASVAB subtests and finishes in a lookup table. Once you can follow the chain, the numbers people quote start to make sense, and so does where your preparation time is best spent.

Step 1: each subtest becomes a standard score

The computer ASVAB, the CAT-ASVAB, adapts as you go. Each subtest starts with a question of average difficulty, updates its estimate of your ability after every answer, and finishes with a final estimate from a three-parameter item response model. Questions you leave unfinished when time runs out are scored as if you had guessed at random.

That estimate is converted to a standard score on a scale with a mean of 50 and a standard deviation of 10 in the 1997 Profile of American Youth, a national sample of about 6,000 people aged 18 to 23. So 50 is average for that group, and the official scores page says about 16% of people score 60 or more on a subtest. Four subtests feed the AFQT: Arithmetic Reasoning (AR), Mathematics Knowledge (MK), Word Knowledge (WK) and Paragraph Comprehension (PC).

Step 2: Word Knowledge and Paragraph Comprehension become one verbal score

WK and PC are not added up. They are combined into a verbal score, VE, which the Defense Manpower Data Center describes as an optimally weighted composite of the two unrounded standard scores. Paragraph Comprehension is weighted at 0.61934 of Word Knowledge, and the result is put back on the same mean 50, standard deviation 10 scale and rounded. One service page describes WK and PC as simply added together; that is a simplification.

Step 3: add them up, counting the verbal score twice

The AFQT sum is AR + MK + 2 × VE, using rounded scores. Because VE counts twice, the verbal half of the battery carries exactly as much weight as both maths subtests together. Push the arithmetic through the verbal formula and one standard-score point on Word Knowledge is worth about 1.3 points of the sum, more than a point on either maths subtest, while a point on Paragraph Comprehension is worth about 0.8.

Step 4: a published table turns the sum into a percentile

The sum is looked up in the conversion table published with the 1997 score scale, in the Defense Manpower Data Center's technical report. A sum of 109 or less is the 1st percentile and 269 or more is the 99th. The official norming page says these norms have been in use since 2004. We found no later document that restates the table, so it is the current conversion as far as the public record shows.

Seven worked examples

Each example runs four standard scores through the formulas above. The last one uses unrounded scores, which is what the verbal formula takes, to show where rounding happens.

Why four average scores give 48, not 50

It feels as if 50 on everything should come out at the 50th percentile. It does not, because the percentile ranks your sum, not your subtests, and in the 1997 youth sample the distribution of that sum is skewed. A sum of 200 sits at the 48th percentile. To reach the 50th you need a sum of 202, one point above average on the verbal score, or two spread across the maths subtests.

Why about 46 on each clears 31

The Army and the Air Force publish an AFQT minimum of 31 for high school graduates, and the table puts 31 at a sum of 183. Standard scores of 45 on all four give 180, the 28th percentile. Scores of 46 on all four give 184, the 32nd. So the line sits between 45 and 46 across the board: less than half a standard deviation below average. Because the verbal score counts twice, a candidate a little under that on maths can make it up on Word Knowledge and Paragraph Comprehension, and the reverse.

Why you will never see 37, 58 or 65

The table is not smooth. A sum of 189 is the 36th percentile and 190 is the 38th; 209 is the 57th and 210 the 59th; 215 is the 64th and 216 the 66th. No sum lands on 37, 58 or 65, so no candidate is ever told those numbers. One consequence: Category II starts at 65 on paper, but the lowest Category II percentile anyone can actually receive is 66.

What the percentile is used for

Your percentile places you in an AFQT category, and the services set their enlistment minimums on it. These are the published figures as of October 2026; confirm yours with a recruiter.

The AFQT only decides whether you can enlist. Which jobs are open depends on line scores and composites, which are built from the same standard scores plus the subtests the AFQT leaves out, such as Mechanical Comprehension and Electronics Information. Our ASVAB hub lists them service by service.

What this means for your preparation

Three things follow from the arithmetic. First, the verbal score is half of the AFQT, so vocabulary and careful reading are not the soft option: a point on Word Knowledge moves the sum more than a point on either maths subtest. Second, Arithmetic Reasoning and Mathematics Knowledge are done without a calculator on every version, so practise them by hand. Third, finish every subtest. Unfinished questions are scored as random guesses, and the official advice warns that several wrong answers at the end of a subtest are heavily penalised, so steady work beats a rush at the end.

Our ASVAB practice tests adapt the way the CAT-ASVAB does and run your answers through these exact formulas and the published table. The questions and their difficulty levels are our own, so treat the result as a guide rather than a prediction. The AFQT Predictor Test practice and Arithmetic Reasoning are free, and the full CAT-ASVAB battery gives you every standard score, your AFQT and the service composites in one sitting.