How Your AFQT Score Is Calculated

Your AFQT score is a percentile from 1 to 99, not a percent correct. It comes from four ASVAB subtests: Arithmetic Reasoning (AR), Mathematics Knowledge (MK), Word Knowledge (WK) and Paragraph Comprehension (PC). Each subtest gets a standard score, WK and PC are combined into a Verbal Expression (VE) score, and the total AR + MK + 2 × VE is looked up in a table built from a 1997 national sample of 18- to 23-year-olds. An AFQT of 60 means that about 60% of that sample scored at or below your total.

This guide walks through each step with the published numbers, works one example, explains why any practice test, ours included, can only estimate the result, and ends with three practice questions.

The four steps at a glance

1. Standard scores. Each subtest is scored on a scale where the 1997 reference group averaged 50, with a standard deviation of 10.
2. Verbal Expression. WK and PC are combined into one verbal score, with PC counting about two-thirds as much as WK.
3. The AFQT composite. AR + MK + 2 × VE. The verbal score counts twice.
4. The percentile. The composite is converted to a percentile from 1 to 99 with a published table.

Step 1: Standard scores

The ASVAB does not simply count right answers. Its scoring model takes account of which questions you answered, how hard each one is, and the chance of guessing right on a multiple-choice question, and it pulls very uncertain scores toward the average. On the computer version (the CAT-ASVAB) the questions also adapt: a right answer usually brings a harder question next, and a wrong answer an easier one. So two people with the same number of right answers can get different scores.

That ability estimate is then put on the standard-score scale. The published equation is for the reference test form the scale was built on; every other paper form and computer item pool is statistically equated to the same scale.

On this scale, 50 is the 1997 group's average and every 10 points is one standard deviation. In the ASVAB program's words, "About half the population scores at or above a Standard Score of 50 and about 16% of the population scores at or above a Standard Score of 60." A standard score is neither a percentile nor a percent correct.

One timing rule belongs here too. On the proctored computer version, if time runs out before you finish a subtest, the questions you did not reach are scored as though you had answered them at random. The ASVAB program says the time limits are generous enough that nearly everyone finishes each subtest, so this is rarely applied. On the paper version, still given at a few MET sites, a question left blank is simply scored wrong, so answer every question. The PiCAT, an unproctored version taken online, has no time limit per subtest.

Step 2: Verbal Expression (VE)

VE is not the sum or the average of your WK and PC scores. When the current score scale was built, VE was defined as a weighted combination in which PC gets about 0.62 of WK's weight (the report calls it "about two-thirds"), rescaled so that VE also averages 50 with a standard deviation of 10 in the reference group. The weights were chosen to match how the older 1980 scale weighted the two tests. There PC in effect counted for less because it is the shorter test, which the report calls desirable given PC's lower reliability.

Written with the two standard scores, the formula is approximately:

The report states this formula in terms of ability estimates (its equation 2.3). The version above is the same equation rewritten with the report's own constants (Table 2.4), with the weights rounded to three decimals and the constant to two. The ASVAB's own scoring also uses the unrounded WK and PC scores at this step, so working from the whole-number scores on a score report can put you a point off.

Some recruiting pages, the Air Force's among them, say WK and PC are "added together" to form VE. That is shorthand. The 1980 scale did add the raw number of right answers on the two tests; the current 1997 scale weights the two standard scores unequally, as described here.

Step 3: The AFQT composite

The composite adds the whole-number standard scores, with VE counted twice:

Half of the composite is verbal and half is math. One extra standard-score point in AR or MK moves the composite by exactly 1. Because WK carries more of VE than PC does, an extra point in WK moves it by about 1.3 on average and an extra point in PC by about 0.8. Those are averages: VE is rounded to a whole number before it is doubled, so any single extra WK or PC point moves the composite by either 0 or 2. That is arithmetic, not study advice: a point is not equally easy to gain on every subtest.

Step 4: From composite to percentile

The composite is converted to a percentile with a table in the same 1997 score-scale report. Each row gives the percentage of the reference group that scored at or below that composite, rounded to a whole number, with 0 reported as 1 and 100 as 99. A few rows:

The table is not a straight line. In the middle, one composite point is worth about one percentile point; near either end it takes several composite points to move one percentile. Because of the rounding, a few percentiles (37, 58 and 65) never appear at all.

The reference group matters as much as the arithmetic. The percentile compares you with a national sample of 18- to 23-year-olds who took the ASVAB in 1997, not with the people who tested on the same day as you.

A worked example

These standard scores are made up for illustration; they are not anyone's real result.

1. VE: 0.655 × 56 + 0.406 × 51 − 3.03 = 36.68 + 20.71 − 3.03 = 54.36, which rounds to 54.
2. Composite: 54 + 52 + 2 × 54 = 214.
3. Percentile: the table puts a composite of 214 at the 63rd percentile, in Category IIIA (50 to 64).

The rounding caveat from Step 2, in action: if the unrounded WK and PC scores behind those whole numbers were 56.4 and 51.4, VE would round to 55 instead, the composite would be 216, and the percentile would be 66, which is Category II (65 to 92) instead of IIIA. Near a category line, one rounding step can change the category.

AFQT categories

AFQT percentiles are also grouped into categories:

Each service sets its own minimum AFQT for enlistment. Minimums can depend on your education credential and change with recruiting needs, and a recruiter's current standard is the one that counts, so this guide does not list them.

Common mix-ups

• "An AFQT of 50 means I got half right." No. It is a percentile: about half of the 1997 reference group scored at or below your composite.
• "Adding my four standard scores gives my AFQT." No. VE weights WK and PC unequally and counts twice, and the total then goes through the percentile table.
• "A standard score of 60 is the 60th percentile." No. A standard score is on the scale where the 1997 reference group averaged 50 with a standard deviation of 10, and about 16% of that group scored 60 or higher. A percentile, like the AFQT, is a different scale. Service job composites use their own scales: the Air Force's are percentiles, the Army's and Marine Corps' average 100, and the Navy's are sums of standard scores. Check which scale a number is on before comparing it.
• "My AFQT is my whole ASVAB score." The AFQT uses four subtests. The services also combine subtests into their own composite scores, often called line scores, to decide which jobs you qualify for.

Why a practice test can only estimate your AFQT

Steps 2 to 4 are fixed arithmetic. Step 1 is the hard part, and no practice test can reproduce it exactly:

• Practice questions are not normed. New ASVAB questions are tried out on examinees before they count, and the score scale was set on the 1997 group. Practice questions, ours included, have never been given to a comparable sample, so turning practice answers into a standard score takes an estimate.
• Short practice sets are noisy. Five or ten questions leave a wide margin either way.
• The computer test adapts; most practice tests do not. The real test picks each question based on your earlier answers, so it pins down your level with fewer questions; a fixed practice set cannot copy that.

TopASVAB's scoring follows the published arithmetic on this page, with two differences. We cap standard scores at 20 and 80, because our estimate has no real precision beyond that range (the 1997 scale itself has no such cap). We also limit the ability estimate behind each score to a fixed range (−3 to +3 on our model's scale); in Mathematics Knowledge that limit is reached first, so our highest MK score is 76, not 80. The first step, from practice answers to an ability estimate, is our own estimate, and every standard score and AFQT figure we show is labeled an estimate. Treat any practice-based AFQT, from any site, as a guide to where to study, not a prediction of your score on test day.

Practice questions

Each question has one best answer. Work it out with the formulas and the table on this page before you open the explanation. The standard scores in these questions are made up for illustration.

Where to go next

• The Arithmetic Reasoning guide covers the AFQT's word-problem subtest.
• The Word Knowledge guide covers the AFQT's vocabulary subtest.
• The ASVAB program's own page on understanding ASVAB scores.
• Free sample questions for every subtest from the ASVAB program.

Sources

• Segall, D. O. (2004). Development and Evaluation of the 1997 ASVAB Score Scale (Technical Report No. 2004-002). Defense Manpower Data Center. Sections 2.2 to 2.4: equations 2.1 to 2.3 and how the 1980 scale weighted WK and PC, Table 2.4 (standard-score constants for the reference form, 04D), the untruncated standard-score range, and Table 2.5 (AFQT percentile conversion). Section 2.5: the services' job composite scales (2.5.1, the Air Force's percentiles; 2.5.2 and 2.5.3, the Marine Corps' and Army's mean-100 scales; footnote 5, the Navy's unscaled sums). PDF on officialasvab.com
• officialasvab.com, ASVAB scores for applicants: the four AFQT subtests, the 1 to 99 percentile, the 1997 reference group of 18- to 23-year-olds, the standard-score quotation in Step 1, and the AFQT categories.
• officialasvab.com, ASVAB scores for researchers: standard scores with a mean of 50 and a standard deviation of 10, scores on the same scale whichever questions you get, equating for every computer item pool and paper form, and unanswered paper-test questions scored as wrong.
• officialasvab.com, CAT-ASVAB: how the computer version adapts, how questions left unfinished are scored, and that nearly all examinees finish each subtest in time.
• officialasvab.com, FAQs: what an AFQT percentile means.
• officialasvab.com, What to expect: tryout questions that do not count toward your score, the paper version given at only a handful of MET sites, and the PiCAT's lack of time limits for individual subtests.
• airforce.com, ASVAB: the Air Force's job composites (Mechanical, Administrative, General Aptitude and Electronics) are reported as percentiles from 1 to 99. The same page describes VE as WK and PC "added together"; Step 2 explains why that is shorthand.
• The practice questions and worked answers were written for this guide by TopASVAB.