Testing

How to Read ADHD Test Scores: Percentiles, T-Scores, and What They Mean

September 14, 2026 · 6 min read

Last updated September 14, 2026

An ADHD report contains tables of numbers that most readers skip. That is a loss, because the numbers are where the evidence is, and they are not hard to read once you know the three conventions almost every instrument uses. This is part two of the testing series.

Percentiles

A percentile tells you where a score falls relative to a comparison group of the same age (and sometimes sex). A 50th percentile is exactly average. A 90th percentile on an inattention scale means the person's reported inattention is higher than 90 percent of peers. On symptom scales, higher percentiles mean more symptoms; on ability tests, higher means stronger. Check which direction the report is using, since a 5th percentile on a working memory test and a 95th percentile on an inattention scale are both describing a weakness.

T-scores

Many rating scales report T-scores: a scale where 50 is average and each 10 points is one standard deviation. A T-score of 60 is roughly the 84th percentile, 65 is about the 93rd, and 70 is about the 98th. Most scales flag 65 or 70 as 'clinically significant' or 'very elevated,' meaning the score is in the range where a diagnosis becomes plausible. A T-score of 60 to 64 is often 'at risk' or 'elevated': notable, not conclusive.

Standard scores and scaled scores

Cognitive tests use standard scores, where 100 is average and 15 points is one standard deviation, so 85 to 115 is the middle two-thirds of people. Subtests often use scaled scores with a mean of 10 and a standard deviation of 3. A working memory index of 85 alongside a verbal index of 115 is a thirty-point gap, and gaps like that are what clinicians look at, more than any single number.

What 'clinically significant' does and does not mean

A clinically significant score means the reported behavior is unusual enough to warrant clinical attention. It does not by itself mean ADHD. Anxious children score in the clinical range on inattention; sleep-deprived adults score there on everything. The clinician's job is to take a set of significant scores, check them against history and the other instruments, and decide what explains them. A report that lists 'clinically significant' scores and stops has not finished.

Reading disagreement between raters

Two people rating the same child usually produce different scores, and the pattern is meaningful. A parent T-score of 72 and a teacher T-score of 55 says the problem is largely at home, and the clinician should ask why. Teacher 70 and parent 58 suggests school demands are exposing something home does not. Both above 65 is the pervasive pattern consistent with ADHD. Self-ratings in adolescents and adults are typically lower than informant ratings, and a self-rating far above a partner's rating is worth a closer look.

Performance test outputs

Continuous performance tests report several numbers: omissions (missed targets, a measure of inattention), commissions (responses to non-targets, a measure of impulsivity), reaction time, and reaction time variability, which is often the most sensitive ADHD marker. Each is reported as a standard score or percentile against age norms, and the report usually gives an overall index. Look at variability and omissions first; then at whether the pattern fits what the scales and interview found.

Validity indicators

Good scales include checks for inconsistent or exaggerated responding. If the report mentions that validity indices were within normal limits, the rater answered consistently. If it flags a validity concern, the clinician should say how they weighed those scores. This matters most in adult evaluations, where the stakes of a diagnosis (medication, accommodations) create pressure in both directions.

The bottom line

Percentiles locate a score against peers; T-scores of 65 or 70 mark clinical significance; standard scores of 100 are average with gaps of 15 or more worth attention. None of these numbers is a diagnosis; the pattern across instruments and raters, read against the history, is. Part three covers what testing costs and why.