What Is a Good IQ Score? Ranges, Percentiles & What It Means

You just took an iq test. The screen says 118, and now you're staring at it, wondering: is that good? Is 118 "smart"? Should it be 125? Did you fail something?

The number sits there with no context, and most explanations online either drown you in statistics or hand out flattering labels like participation trophies. The honest answer: what counts as a "good" IQ score depends on the scale, the margin of error behind your number, and — most importantly — what you need the score to do. Let's take all three in order.

The Short Answer: What Counts as a Good IQ Score?

If you want the quick reference before the explanations, this is it. On virtually all modern IQ tests, the average score is 100 and the standard classification breaks down like this:

IQ Range	Classification	What It Means
130 and above	Very superior / Gifted	Top 2% — the Mensa threshold
120–129	Superior	Top 10% — typical of many university graduates
110–119	High average	Clearly above average
90–109	Average	Where roughly half the population lands
80–89	Low average	Below the midpoint, still within the normal range
70–79	Borderline	May warrant support in academic settings
Below 70	Extremely low	Clinically significant only with adaptive-function assessment

So a score of 120 is genuinely good — better than roughly nine out of ten people. A score of 110 is solidly above average. And 100 is exactly average, which is not an insult; it's the midpoint of the entire population by design.

Now the details, because the scale has some surprising mechanics.

Why 100 Is the Average (and What Standard Deviation Means)

The original "IQ" stood for intelligence quotient — literally a quotient. Early tests in the 1910s divided your mental age by your chronological age and multiplied by 100. A ten-year-old performing like an average twelve-year-old got an IQ of 120. That formula is obsolete; it breaks down for adults and was abandoned decades ago.

Modern tests use what psychometricians call a deviation IQ. Your raw performance is compared against a large, representative sample of people your age — the norm group — and your score expresses where you fall in that distribution. The average of the norm group is set to exactly 100, and one standard deviation is set to exactly 15 points.

Because cognitive ability is distributed roughly normally, that gives you the famous bell curve:

• About 68% of people score between 85 and 115 (within one standard deviation).
• About 95% score between 70 and 130 (within two standard deviations).
• Scores beyond that are rare — only about 1 in 44 people scores 130 or above.

The key insight most people miss: 100 is not a fixed amount of intelligence. It's the center of a relative ranking. Your score tells you where you stand compared to other people — not how many questions you answered correctly, and not some absolute quantity of "smart."

The Full IQ Score Classification Table (With Percentiles)

Labels are convenient, but percentiles are more intuitive. A percentile tells you the percentage of the population you scored at or above. The full mapping on the standard scale (mean 100, SD 15):

IQ Score	Percentile	How Rare Is That?
145	99.7th	About 1 in 400
130	98th	About 1 in 44
125	95th	Top 1 in 20
120	91st	Top 1 in 11
115	84th	Top 1 in 6
110	75th	Top 1 in 4
100	50th	Exactly the midpoint
85	16th	Bottom 1 in 6
70	2nd	Bottom 1 in 50

A few translations. If you scored 115, you outperformed about five out of every six people — "high average" honestly undersells how uncommon that is. If you scored 130, you're at the threshold Mensa, the high-IQ society, uses for admission. And if you scored 90, you're within shouting distance of half the planet; there is no meaningful problem to solve.

One caution: percentile scales are stretched at the edges. The gap between the 50th and 60th percentile is a small ability difference; the gap between the 98th and 99.9th is enormous. Percentiles compress the middle and amplify the tails, which is why test reports use scores rather than percentiles for fine distinctions.

Your Score Is a Range, Not a Point

The fact most IQ score articles skip, and it changes how you should read your number: no psychological test measures with perfect precision.

Every score carries measurement error, quantified as the standard error of measurement (SEM). The math is simple: SEM = 15 × √(1 − reliability). A well-designed professional test with a reliability of .95 has an SEM of about 3.4 points, which means your true score is most likely within roughly ±7 points of the reported one at 95% confidence.

The consequences:

• Someone who scored 118 and someone who scored 124 are statistically indistinguishable. Their confidence intervals overlap almost completely.
• Retake the same good test a week later and expect your score to wander by several points even though nothing about you changed. A five-point bounce is noise, not growth or decline.
• A report that gives you "132" with no confidence interval is selling false precision. The testing standards published jointly by AERA, APA, and NCME treat the score interval, not the point, as the meaningful unit.

Read your result as "somewhere in the low 120s," not "120.000." The band is the honest answer.

What "Good" Means Depends on Why You're Asking

"Good relative to what?" is the question the classification table can't answer for you. Four contexts, four different answers.

For everyday life. A score in the average range or above is sufficient for nearly every job, degree, and daily cognitive demand. Research on life outcomes consistently shows that above roughly 120, extra points deliver diminishing returns — past that level, motivation, conscientiousness, and opportunity do far more work than additional IQ points.

For jobs and academics. Cognitive ability is one of the best single predictors of job performance — meta-analytic work by Schmidt and Hunter found a corrected correlation around 0.5 across jobs — but the prediction strengthens as job complexity rises. Low-complexity roles show little benefit from above-average scores; highly technical and academic tracks effectively filter for the top 10–15%. If your goal is a specific career, "good" means good enough for that track, which is usually a much lower bar than 130.

For clinical purposes. Scores below 70, combined with deficits in adaptive functioning and developmental onset, are one criterion for an intellectual disability diagnosis under DSM-5. Note the wording: one criterion. No diagnosis is ever made from a number alone. Clinicians also assess daily living skills, developmental history, and functioning across settings — and the practical threshold for independent living sits around 70–75, not at a round number.

For high-IQ societies. Mensa accepts the top 2%, which on an SD-15 scale means 130 or above. That's a membership rule, not a scientific boundary — nothing magical happens at 129.

Why Online IQ Test Scores Often Run High

If you took a free test online and scored 132, an uncomfortable statistic: analyses of popular free online IQ tests routinely find average reported scores of 120–130 — on a scale where the true average is 100 by definition. If most users of a test score 125+, that test is not measuring your IQ; it's flattering you, usually to sell a "premium report."

No regulation covers the term "IQ test." Anyone can build a 15-question quiz, call it an IQ test, and print whatever number converts best. Red flags to check before trusting any online score:

• The score is revealed only after payment. The test was designed to sell you a number.
• No age norming. Without comparison to people your age, a percentage of correct answers is not an IQ score.
• Under 20 questions, or under 10 minutes. A stable estimate needs enough items across enough cognitive domains.
• No standard deviation or confidence interval mentioned. Any real score has error bars; hiding them means they weren't computed.
• Average results cluster around 120–130. Honestly normed tests produce averages near 100.

A well-constructed online assessment — age-normed, multi-domain, transparent about methodology — can land within 5–8 points of a professionally administered test like the WAIS. That's good enough for curiosity and self-insight. It is not good enough for a diagnosis, a gifted-program placement, or any legal purpose; those require an in-person evaluation by a licensed psychologist.

Can You Change Your Score? Practice, the Flynn Effect, and Honest Limits

Three things genuinely move IQ scores, and two of them are misunderstood.

Practice effects are real but narrow. Familiarity with question formats reliably improves performance — mostly by stopping the format itself from costing you points. Walking into a matrix-reasoning section already knowing how rotation, distribution, and overlay rules work means you spend your minutes reasoning, not decoding instructions. That's why working through the common IQ test question types is the highest-leverage preparation there is. What practice does not do, on current evidence, is raise the underlying ability the test is sampling.

The Flynn effect is about populations, not individuals. Average IQ scores rose roughly 3 points per decade through most of the 20th century — likely driven by nutrition, education, and increasing exposure to abstract, analytical thinking. Some countries now show the effect reversing. Either way, this happens across generations; it is not a lever you can pull before next Tuesday's test.

Your state on test day matters more than people admit. Sleep deprivation, test anxiety, a noisy room, or a poor night's sleep can each move a score by several points. A "bad day" score underestimates you; an anxious retake can overestimate the decline. One more reason to treat a single number as a range.

And the honest limit: claims that brain training raises general intelligence remain contested. Controlled studies of working-memory training show reliable improvement on the trained tasks, but transfer to overall IQ scores is small and disputed. Training is still worthwhile — for attention, processing speed, and the specific skills games exercise — just not as an IQ-inflation scheme.

The Bottom Line

Three takeaways, in order of importance:

1. A good score is a range, and its meaning is contextual. 90–109 is average; 110–119 is above average; 120+ puts you in the top 10%; 130+ in the top 2%. Any of those is "good" for the purposes they need to serve.
2. Your exact number is softer than it looks. With a ±7-point confidence band on even professional tests, a 118 and a 124 are the same answer given twice.
3. Use scores as a starting point, not a verdict. IQ is one input among many — EQ predicts real-world outcomes at least as strongly, and neither number caps what practice and persistence do.

If this article started because you're preparing for a test, don't stop at the score scale — learn the question formats themselves, because format familiarity is the most reliable score-protector there is. And if you want to train the cognitive muscles tests actually sample — working memory, attention, pattern recognition — start with a few rounds of memory match on iQiQGame. The score is a snapshot; the skills are trainable.

References

• Wechsler classification and deviation IQ: en.wikipedia.org/wiki/Wechsler_Adult_Intelligence_Scale
• Schmidt & Hunter (1998), "The Validity and Utility of Selection Methods in Personnel Psychology": psycnet.apa.org/record/1998-10687-006
• DSM-5 intellectual disability criteria: psychiatry.org
• AERA/APA/NCME testing standards context: ncme.org
• Flynn effect overview: en.wikipedia.org/wiki/Flynn_effect