Leiter-3 Scores Explained: Nonverbal IQ, Composites and Classification Ranges

The Leiter-3 reports a Nonverbal IQ on the usual mean-100, standard-deviation-15 scale, from 30 to 170, and every part of it is given without spoken instructions. The score is the sum of the scaled scores on four cognitive subtests, and the sum maps to an IQ through a table in the manual: four scaled scores of 10, the average for each subtest, give 98, and a sum of 42 gives 100. The test is built for people who are hard to test in words, such as autistic children, children with speech or language disorders, deaf or hard-of-hearing children and those who do not speak English. This guide sets out the subtests, how the IQ and the two memory and speed composites are scored, the manual's classification labels, and what independent research says about how far the scores can be trusted.

What is the Leiter-3, and who is it for?

The Leiter International Performance Scale, Third Edition was published by Stoelting in 2013 and written by Gale Roid, Lucy Miller, Mark Pomplun and Chris Koch. It covers ages 3 to 75 and older. Stoelting gives 20 to 45 minutes for the cognitive scales and 30 minutes for the separate Attention/Memory battery. The publisher describes it as administered completely nonverbally: examiners use a gestural card and pantomime, with teaching trials (up to three tries) on easy items, and the child responds by placing blocks in a frame. It is a restricted test: Stoelting classes it at user level B, which requires graduate coursework in tests and measurement, and the kit cost about 1,600 dollars when we read the listing. It is not something a parent can take online.

The Leiter-3 shortened its predecessor. Stoelting says the number of subtests was cut from 20 in the Leiter-R to 10. The manual reports special-group data for ADHD, autism, severe cognitive delay, English-as-a-second-language, gifted and talented, severe hearing impairment, learning disability, severe motor delay and severe speech or language impairment. For the older nonverbal tests our site covers, see Raven's matrices and the Cattell Culture Fair test, and for the blog overview of tests like this one, professional IQ tests.

What are the ten Leiter-3 subtests?

• Cognitive battery: Figure Ground, Form Completion, Classification/Analogies and Sequential Order feed the Nonverbal IQ
• Optional cognitive subtest: Visual Patterns, which can replace one of the four core subtests (Stoelting's web page names it Matching/Repeated Patterns; the manual and record form say Visual Patterns)
• Attention/Memory battery: Forward Memory and Reverse Memory feed the Nonverbal Memory composite
• Attention/Memory battery: Attention Sustained and Nonverbal Stroop (incongruent correct) feed the Processing Speed composite
• Attention Divided is the fifth Attention/Memory subtest; the appendices we read do not say which composite, if any, it feeds

Note what is not there. The Attention/Memory battery yields exactly two composites, Nonverbal Memory and Processing Speed, and no attention composite. Working memory and processing speed are also left out of the IQ on purpose: the Nonverbal IQ rests on the four cognitive subtests only, which is a design choice the Wechsler scales do not make. Compare it with how the full-scale IQ is built from index scores on the Wechsler scales.

How is the Leiter-3 Nonverbal IQ scored?

Each cognitive subtest yields a scaled score with a mean of 10 and a standard deviation of 3, on a range of 0 to 20. The four scaled scores are added, and the sum is looked up in the manual's Appendix D.1. The table runs from a sum of 0 (IQ 30) to a sum of 80 (IQ 170). Selected rows from Stoelting's published appendix:

• Sum 8 gives IQ 39; sum 24 gives 70; sum 30 gives 77; sum 32 gives 81; sum 35 gives 90
• Sum 38 gives 96; sum 40 gives 98; sum 42 gives 100; sum 44 gives 103
• Sum 50 gives 113; sum 60 gives 126; sum 70 gives 143; sum 73 gives 149; sum 74 gives 152; sum 80 gives 170

Two things follow from the table. A child who scores an average 10 on all four subtests gets 98, two points below the mean, because the sum needed for 100 is 42; the table is built to the norm sample's actual distribution, not by a formula. And in the middle of the range one scaled-score point on the sum moves the IQ by roughly two and a half points (38 to 96, 40 to 98, 42 to 100, 44 to 103, 50 to 113), so a single subtest that is three points off changes the IQ by about eight. That second point is our reading of the table, not a manual statement. The manual's 95% confidence band for the Nonverbal IQ is plus or minus 6 points in the table, from a standard error of measurement it gives as averaging 3.0 (twice the SEM is the band). Our guide to the margin of error on an IQ score explains why a single result is a band.

What are the Leiter-3 classification ranges?

The manual's Figure 4.2 gives descriptive labels for Nonverbal IQ scores. Stoelting corrected one row after publication: copies printed before the correction and many score charts online show the Very High/Gifted band as 130 to 170, when it is 130 to 149. The corrected table, with the percentiles the manual prints:

• 150 to 170: Extremely High/Gifted (99.9th percentile)
• 130 to 149: Very High/Gifted (98th to 99th)
• 120 to 129: High (91st to 97th)
• 110 to 119: Above Average (75th to 90th)
• 90 to 109: Average (25th to 74th)
• 80 to 89: Below Average (9th to 24th)
• 70 to 79: Low (3rd to 8th)
• 55 to 69: Very Low and Mild Delay (1st to 2nd)
• 40 to 54: Moderate Delay (0.10th)
• 30 to 39: Severe Delay (0.04th)

The labels are not Wechsler's: there is no Borderline, Superior or Very Superior, and a score of 70 to 79 is Low. A reader comparing a Leiter-3 report with a Wechsler report should compare scores and percentiles, not labels. The same labels apply to the Nonverbal Memory and Processing Speed composites. For the Wechsler wording, see our guides to WISC-V index scores and DAS-II scores, and the IQ bell curve for what the percentiles mean.

What do the Nonverbal Memory and Processing Speed scores mean?

Both composites sit on the same mean-100, SD-15 metric, but the manual states the limit in plain words: the composite scores are not equated with IQ scores, because they all measure memory or attentional processes, not global intellectual ability. Nonverbal Memory is the sum of the Forward and Reverse Memory scaled scores and runs from 48 (sum 0) to 153 (sum 40); a sum of 20 is 100. Processing Speed adds Attention Sustained and Nonverbal Stroop and runs from 48 to 156, with 100 again at a sum of 20. The manual's sample report shows a Nonverbal Memory sum of 19 giving 97 (42nd percentile, band 87 to 107) and a Processing Speed sum of 16 giving 90 (25th percentile, band 82 to 96).

Growth scores are a third kind of number. They are on a separate scale that the manual compares to marking a child's height on a wall chart across time, and they ignore age-based standard scores. They show change, not standing among peers, and they are not IQ-type scores. In February 2026 Stoelting published a notice that its scoring software had computed the Nonverbal Memory composite growth score wrongly and had reversed two Stroop error fields, adding that all other scores and composites in the software were unaffected.

How accurate is the Leiter-3 for autism, deafness and English learners?

The evidence is thinner than the test's reputation, and the manual chapters on reliability and validity are not published online, so we could not check its reliability coefficients or its correlations with the Wechsler and Stanford-Binet tests. What can be read is this. For deaf and hard-of-hearing people, the manual reports a clinical study of 46 individuals whose scores did not differ significantly from the normative sample, and Stoelting says the result should not be cited as deaf norms. A peer-reviewed scorecard of nonverbal tests for deaf and hard-of-hearing children (Reesman and colleagues, 2014) marked the Leiter-3 yes on administration guidance and no on published empirical studies of those children.

For autism the picture is mixed. A study of the Leiter-3 against the WISC-IV by Giofre and colleagues (2019) found the two batteries' IQs strongly correlated, and concluded that the WISC-IV might underestimate the potential of children with autism, particularly those with a low-functioning profile; we read the abstract only, so we cannot give the sample size or the gap. A 2026 study of 44 autistic children aged 4 to 12 with limited expressive language found a mean Leiter-3 Nonverbal IQ of 92.45 (SD 13.35), against 69.57 on a verbal quotient the authors derived from the PPVT-III, and only the nonverbal score predicted adaptive functioning two to three years later. The gap resembles one seen in the older Leiter-R, which scored 20.91 points higher than the Stanford-Binet 5 in 47 young children with autism (2013); that study was of the Leiter-R, not the Leiter-3. Our own pages on the other test, Stanford-Binet 5 scores, show how it is scored.

Nonverbal does not mean that language plays no part in the thinking. An independent coding of language demands (Lancaster and colleagues, 2025) found the Leiter-3 uses no verbal instructions but has high potential for verbal strategy use, a different profile from Raven's matrices (many spoken instructions, little room for verbal strategies) or the TONI, and concluded that Raven's and the Leiter-3 are better choices when language poses a challenge. For second-language test-takers more generally, see our guide to second language and IQ test scores.

What changed in the Leiter-3 tables after publication?

In June 2014 Roid wrote that some oddities in the data and models had produced tables that did not adequately estimate the score distribution for the lower IQ range, and that a few individuals might have achieved higher than anticipated scores at the lower end. The low end of the Nonverbal IQ table was rewritten. The manual's case studies show the effect: Amanda's Nonverbal IQ went from 78 to 69, Max, an autistic child with severe speech delay, from 79 to 70, and Randy from 85 to 78. A different table links the two editions: for the same child, a Leiter-R Brief IQ of 70, 100 and 150 corresponds to a Leiter-3 Nonverbal IQ of 74, 103 and 148, from a sample of 60. The manual also warns that the Leiter-3 IQ should not be interpreted as a measure of innate capacity or maximum potential, even though Stoelting's product page calls fluid intelligence the truest measure of innate cognitive ability. For another test that reports memory and speed apart from its IQ, see RIAS-2 scores explained.

• Check which edition and which table version the report used: the low-end table changed in 2014
• Compare scores and percentiles across tests, not labels: Leiter-3 has no Borderline band
• Read a Nonverbal IQ as a band of about plus or minus 6, and a composite as a memory or speed score
• Do not read a Leiter-3 IQ as a prediction: the manual says not to predict a young child's future abilities without overwhelming evidence
• Ask for the norm sample: Stoelting reports 1,603 examinees in the standardization sample; we found no published norm year

To see where a result lands on the same curve, take the IQ Metrics test and use the IQ percentile calculator. For the difference between verbal and nonverbal scores, our blog explains verbal and non-verbal IQ scores.

Common questions