Score Interpretation
All 7 bands, exact percentiles & what they actually mean
IQ score ranges are the standardised bands that translate a raw test score into a meaningful position within a reference population. There are seven standard ranges, defined by the Wechsler Adult Intelligence Scale (WAIS-V) classification system — from Extremely Low (below 70) to Very Superior (130 and above). This guide covers every band with exact percentile data, population shares, occupational context, and practical interpretation notes grounded in the most current psychometric research.
Interactive Tool
Enter any score from 40 to 200 to see its WAIS-V classification, calculated percentile, and interpretation context.
7
Standard bands
WAIS-V classification system
90–109
Average range
Covers ≈50% of the population
SD = 15
Wechsler scale
Mean 100, SD 15 on most modern tests
r = 0.96
Test–retest reliability
WAIS-IV, 298 participants, 22-day interval
IQ score ranges are not arbitrary bands — they are derived from the statistical properties of the normal distribution that IQ scores are calibrated to follow. Every major standardised test sets 100 as the midpoint of the reference population and uses a standard deviation (SD) of 15. The named ranges are simply interpretable segments of that distribution. The most widely used classification system comes from the Wechsler Adult Intelligence Scale (WAIS-V, updated 2024), published by Pearson Assessments.
IQ Distribution — Normal Curve (mean 100, SD 15)
The bell curve shape means that scores cluster densely near 100 and become progressively rarer further out. This clustering also means that small raw score differences near the centre of the distribution correspond to larger percentile jumps than the same point difference at the extremes. Moving from IQ 95 to IQ 105 crosses more of the population than moving from IQ 125 to IQ 135.
All IQ score ranges are norm-relative, not absolute measures. A score of 110 falls in the High Average range only relative to the population used for norming that specific test. A different test normed on a different population or in a different decade could place the same raw performance in a different band. For context on how norming affects what the midpoint of 100 represents, see the average IQ score guide.
The WAIS-V, released in 2024 with updated US norms, typically yields scores approximately 3–5 points lower than the WAIS-IV for the same individual. This is not because anyone got less intelligent — it reflects updated reference population data and the ongoing calibration of the 100-point midpoint. If you compare scores across editions, this difference matters. For a historical perspective on why norms shift over time, see the Flynn Effect section below.
The bar below shows how the reference population is distributed across each range. Width is proportional to the percentage of people in each band. Expand any band for percentile context, SD position, and interpretation notes.
Based on a standard normal distribution (mean 100, SD 15). Actual values vary by test and norm sample.
≈2% of the reference population
Two or more standard deviations above the mean. The WAIS-V labels this range Very Superior. Roughly 1 in 50 people score here on a properly normed assessment. This is the threshold used by most high-IQ societies, including Mensa International — though formal membership requires a supervised clinical test, not an online result. Score variance is proportionally larger at the extremes, so a 5-point swing near 130 is less statistically meaningful than the same difference near the centre of the distribution.
≈7% of the reference population
Clearly above-average reasoning performance. A score here means outperforming roughly 9 in 10 people in the comparison group. NLSY79 occupational data places physicians and surgeons at a mean IQ of approximately 123.7 — within this band. Improvement within the Superior range is mostly about eliminating specific error types, such as rushing spatial rotation items or misreading inductive patterns in later questions, rather than wholesale strategy changes.
≈16% of the reference population
What most people informally call a strong score. NLSY79 research places average college graduate IQ around 115, squarely in this band. Consistent above-median performance without reaching the rarer Superior classification. Most people in this range benefit most from targeted practice on specific item types they find hardest — typically spatial rotation and inductive matrix sequences — rather than general preparation.
≈50% of the reference population
The clinical average band as defined by WAIS-V — the broadest single classification, capturing roughly half the reference population. A score here is solidly within the normal range by every major clinical or research standard. The gap between 90 and 109 represents meaningful variation: a score of 95 and a score of 105 map to quite different percentile positions despite sharing the same band label. Most deliberate practice improvements happen within or just above this range.
≈16% of the reference population
Scores in this band indicate performance below the median but within the lower portion of the normal distribution. Roughly 1 in 6 people score here. Poor test conditions, unfamiliarity with matrix formats, or high time pressure can produce scores in this band even for people whose actual reasoning baseline sits higher. A clean re-test under better conditions — quiet room, well-rested, no distractions — is worthwhile before drawing any conclusions.
≈7% of the reference population
The borderline classification is used in clinical settings as a threshold marker, but on online assessments a score in this range most often reflects testing conditions, format unfamiliarity, or significant distraction rather than a stable cognitive baseline. Multiple sessions under controlled conditions are necessary before treating this result as indicative of your actual performance level.
≈2% of the reference population
Clinical tests apply the Extremely Low classification below IQ 70. The DSM-5 uses this threshold as one component of intellectual disability assessment, alongside adaptive functioning measures. For online IQ-style assessments, a score at this level almost certainly reflects session conditions — significant technical problems, major interruptions, or severe time mismanagement — rather than a stable cognitive baseline.
The table below maps specific IQ scores to their exact percentile position, standard deviation distance from the mean, WAIS-V classification band, and approximate share of the reference population. Values are derived from a standard normal distribution (mean 100, SD 15) and match clinically reported WAIS-V data.
| IQ Score | Std. Deviation | Percentile | Band | % of Population |
|---|---|---|---|---|
| 145+ | +3.0 | 99.9th | Very Superior | <0.1% |
| 140 | +2.7 | 99.6th | Very Superior | <0.5% |
| 135 | +2.3 | 99th | Very Superior | ≈1% |
| 130 | +2.0 | 98th | Very Superior | ≈2% |
| 125 | +1.7 | 95th | Superior | ≈5% |
| 120 | +1.3 | 91st | Superior | ≈9% |
| 115 | +1.0 | 84th | High Average | ≈16% |
| 110 | +0.7 | 75th | High Average | ≈25% |
| 105 | +0.3 | 63rd | Average | ≈37% |
| 100 | 0 | 50th | Average | 50% |
| 95 | −0.3 | 37th | Average | ≈37% |
| 90 | −0.7 | 25th | Average | ≈25% |
| 85 | −1.0 | 16th | Low Average | ≈16% |
| 80 | −1.3 | 9th | Low Average | ≈9% |
| 75 | −1.7 | 5th | Borderline | ≈5% |
| 70 | −2.0 | 2nd | Borderline | ≈2% |
| 65 | −2.3 | 1st | Extremely Low | ≈1% |
| 55 | −3.0 | 0.1st | Extremely Low | <0.1% |
Highlighted rows mark key SD boundaries (IQ 70, 100, 130). Classification labels follow the WAIS-V descriptive system.
The percentile column tells you what share of the comparison population scored below a given IQ. A score at the 84th percentile means you outperformed 84% of the normed reference group — it says nothing about the absolute number of items answered correctly or the difficulty of the test. Percentile is a relative position, not an absolute measure of knowledge or skill. For a deeper walkthrough, see the IQ score chart guide, which explains the bell curve and percentile mechanics in detail.
A raw score of 90 and a raw score of 109 both carry the Average label, but they sit at the 25th and 73rd percentile respectively — a gap of 48 percentile points. Two people in the same band can have very different population positions. Whenever possible, refer to the specific percentile your test reports rather than relying on the descriptive band name alone. For high-end scores, the high IQ score interpretation guide covers the 110, 120, and 130+ bands with additional context.
Not all IQ score ranges work on the same numerical scale. The three most commonly referenced scales — the Wechsler (WAIS-V), the Stanford-Binet 5, and the Cattell Culture Fair Intelligence Test (CFIT) — share a mean of 100 but differ in standard deviation. This difference changes what a given number means.
| Feature | WAIS-V | Stanford-Binet 5 | Cattell CFIT |
|---|---|---|---|
| Standard Deviation | 15 | 15 | 24 (original CFIT) / 16 (newer forms) |
| Mean | 100 | 100 | 100 |
| Age Range | 16–90 | 2–85+ | Adult |
| 98th percentile score | 130 | 130 | 148 (SD-24) / 132 (SD-16) |
| Latest edition | WAIS-V (2024) | SB5 (2003, with 2023 updates) | CFIT Scale 3 (various editions) |
| Primary use context | Clinical adult assessment | Clinical lifespan assessment | Culture-reduced fluid reasoning |
Since WAIS-V and Stanford-Binet 5 both use SD = 15, their scores are numerically equivalent at the same percentile. A score of 130 on either test corresponds to the 98th percentile. The Cattell CFIT (original, SD = 24) does not align numerically: a score of 148 on the Cattell equals approximately 130 on the Wechsler — both are at the 98th percentile. This is a common source of confusion when reading quoted IQ scores online. The key is always the percentile, not the number itself.
IQMog reports scores on a standard SD-15 scale (mean 100), aligned with the WAIS-V and Stanford-Binet 5 convention. The classification band labels follow the WAIS-V descriptive system. Your score and percentile are directly comparable to those systems at the same percentile position. For context on how Mensa eligibility thresholds map to these scales, see the Mensa IQ score requirements guide.
When you see celebrity or historical IQ claims cited online — frequently in the 160–200 range — they are almost always unverified estimates. No standardised test has sufficient ceiling to measure scores above roughly 160–170 with statistical reliability. Scores above 200 are physically impossible on any normed instrument. The APA’s intelligence overview explains the appropriate use context for psychometric assessment.
Tests normed in earlier decades — WAIS-I through WAIS-IV — each used their own reference population. Because average raw performance has risen over time (the Flynn Effect, described below), a raw score that earned IQ 105 on the WAIS-IV norms might earn IQ 100 on the WAIS-V. Comparing scores across test editions requires accounting for this shift.
An IQ score range tells you where a single session’s performance sat relative to a normed comparison group. It does not tell you about creativity, emotional intelligence, practical problem-solving, domain expertise, or any of the other dimensions that shape real-world outcomes. Ranges are orientation points — a rough map, not a verdict.
The National Longitudinal Survey of Youth (NLSY79, n = 9,774) provides the most widely cited occupational IQ averages. These are population means, not individual cutoffs. Within-occupation standard deviations of 10–15 points mean any two adjacent occupations overlap substantially in their IQ distributions.
| Occupation | Mean IQ (NLSY79) | Within-occupation SD |
|---|---|---|
| Physician / Surgeon | 123.7 | 10.3 |
| Computer Programmer | 116.2 | 12.1 |
| Computer Systems Analyst | 111.1 | — |
| Electrical Engineer | 105.8 | — |
| Registered Nurse | 104.0 | — |
| Police Officer | 99.4 | — |
| Office Supervisor | 96.8 | — |
Source: NLSY79 dataset analysis. Occupational means are population averages; individual variation within each occupation is wide.
Meta-analytic research places the correlation between IQ and academic grades at approximately r = 0.54 across all educational levels (Roth et al., 2015, k = 240 studies). The correlation between IQ and job performance ranges from r = 0.31 for all occupations to r = 0.51 for cognitively complex roles (Schmidt & Hunter, 1998, updated analyses). These are meaningful population-level correlations, but they leave vast room for individual variation.
A correlation of r = 0.54 between IQ and grades means roughly 29% of grade variance can be statistically attributed to IQ differences across the population. The remaining 71% reflects everything else: motivation, study habits, teaching quality, interest, and circumstances. At the individual level, a single IQ score range predicts very little about any specific outcome with certainty. The NIH StatPearls review on intellectual disability offers additional context on how clinical thresholds are applied — and why adaptive functioning matters as much as score range in formal diagnosis.
IQ norms are not fixed. The Flynn Effect — named after researcher James Flynn, who documented the trend in 1984 — describes the sustained rise in raw IQ scores across many countries throughout the 20th century. Because tests are periodically re-normed against current populations, this means the same raw score that earned IQ 105 in 1980 might earn IQ 100 on updated norms today.
A 2014 meta-analysis (Trahan et al., PMC4152423) found a mean gain of approximately 2.93 IQ points per decade across multiple countries from the 1930s to the 1990s. Fluid reasoning tests (like Raven’s Progressive Matrices) showed gains of up to 5–6 points per decade. Researchers attribute most of the gain to improved education access, better nutrition, reduced childhood lead exposure, and increasing familiarity with abstract test formats.
After roughly 70 years of gains, the Flynn Effect has stalled or slightly reversed in several high-income countries. Bratsberg & Rogeberg (2018, PNAS) documented a reversal in Norway using within-family data, demonstrating the change is environmentally caused rather than genetic. Denmark showed similar patterns: gains of 3 points/decade from 1959–1979 declined to approximately −1.5 points/decade by the early 2000s. The UK, Finland, France, and Australia have shown comparable stagnation or modest declines since the mid-1990s.
A 2024 study tracking a German-speaking cohort from 2005–2024 found gains continuing for initially lower-scoring individuals but weakening overall coherence between cognitive abilities. The WAIS-V re-norming (Pearson, 2024) confirms the historical pattern: the same performance that would have produced IQ 105 on WAIS-IV norms produces approximately IQ 100–102 on WAIS-V norms, reflecting adjusted reference population data. For score interpretation, the practical implication is to verify which norm sample your test uses before comparing across editions.
Online IQ-style tests can produce useful range estimates, but they operate under different conditions than clinical instruments and their ranges carry different levels of precision. Understanding these differences helps you calibrate how much weight to place on your online result.
The ICAR-16, a freely available validated cognitive battery, achieves a correlation of approximately r = 0.80 with the WAIS-IV. The 60-item ICAR version reaches r = 0.85–0.90. By comparison, two administrations of the WAIS itself correlate at r = 0.96. Free short-form “IQ tests” with undisclosed norming and no published validation data may correlate as low as r = 0.20–0.40 with genuine IQ. The quality gap is wide, and the key is always the norming methodology and validation evidence behind a specific instrument.
What good norming looks like:
Treat any online result as a directional estimate rather than a precise placement when:
In any of these cases, one additional clean-session attempt — quiet room, rested, no interruptions, familiar format — will give you a more reliable picture of your actual range placement. IQMog uses a fixed Raven-style dataset and consistent scoring model specifically to reduce session-to-session variance compared to tests that randomise items.
The most productive use of a score range is as a starting point for deliberate practice, not as a verdict on your cognitive ceiling. Ranges tell you roughly where you are — they do not tell you where you are capped.
Most errors at this level come from rushing the final items under time pressure or from spending too long on early items. Fix your pacing first before any other change.
Work specifically on spatial rotation items and inductive matrix sequences, where item difficulty spikes sharply. General practice is less effective than targeted work on your weakest item types.
A single recurring error type — like misidentifying rotation direction or missing a pattern rule on complex matrices — can suppress an otherwise high result. Identify and eliminate your specific error categories.
A range placement is most reliable when it appears consistently across two or more sessions under comparable conditions. A single strong session might reflect optimal conditions rather than a stable baseline. A single weak session might reflect distraction rather than actual performance. Two consistent results give you meaningful signal about your genuine range.
The single biggest lever on your next result is environment. A quiet, distraction-free session at a consistent time of day when alertness is at its peak will produce a more accurate range placement than any amount of practice under poor conditions. Get your environment right before drawing conclusions from any single result.
The WAIS-V defines seven standard IQ score ranges: Extremely Low (below 70, below 2nd percentile), Borderline (70–79, 2nd–8th percentile), Low Average (80–89, 9th–24th percentile), Average (90–109, 25th–73rd percentile), High Average (110–119, 75th–90th percentile), Superior (120–129, 91st–97th percentile), and Very Superior (130 and above, 98th percentile and above). These bands assume a mean of 100 and a standard deviation of 15.
The clinical average band runs from 90 to 109, capturing roughly 50% of the normed reference population and spanning the 25th to 73rd percentile. The broader statistical average — one standard deviation either side of 100 — stretches from 85 to 115 and includes approximately 68% of the population. Any score in either of these ranges is solidly within the normal range by every major clinical and research standard.
Scores of 110–119 (High Average, top 10–25%) are widely considered above average. Scores of 120–129 (Superior, roughly top 3–9%) represent clearly elevated reasoning performance. Scores of 130 and above (Very Superior, top 2%) are the threshold for most high-IQ societies, including Mensa International. These thresholds apply to SD-15 scales such as the WAIS-V and Stanford-Binet 5.
Your range position can shift based on testing conditions, format familiarity, stress, and which specific test or norm sample is used. WAIS-IV test–retest studies with a correlation of r = 0.96 show an average variance of fewer than 5 points over roughly 22 days, but individual sessions can show swings of 10–15 points or more under suboptimal conditions. Treat any single result as a directional estimate within a range, not a fixed label. Consistent results across two or more controlled sessions provide more reliable evidence of your actual band placement.
Well-designed online IQ tests using validated item sets correlate at r = 0.60–0.80 with clinically administered instruments like the WAIS-V. Free short-form tests with undisclosed norming may correlate as low as r = 0.20–0.40. IQMog uses a fixed Raven-style item dataset and a consistent scoring model, which reduces session-to-session variance, but like any online instrument it is best treated as a directional reasoning baseline rather than a clinically precise measurement.
Next Step
Free to start and complete. Raven-style matrix reasoning test, 20 questions, no email required to begin. Unlock your IQ estimate, percentile band, and detailed cognitive profile after the test for $2.99. Built on a fixed item dataset and consistent scoring model for repeatable, comparable results.
Not for clinical diagnosis • Not for Mensa admission • For personal benchmarking only