Scoring Mechanics

IQ 100: the exact middle of the scale, and what follows from it

An IQ of 100 is the one score that is true by definition: the mean of the norm sample, the 50th percentile, and the centre of the Average band from 90 to 109. This page explains why the number is 100, how wide the middle is, why a 100 is an interval rather than a point, how it changes with the scale and the norm year, and what an average score does and does not predict.

A dense crowd of people seen from behind, walking down a sunlit city street, their faces out of focus.
An IQ of 100 is the middle of a crowd like this one, with half of the reference group ahead and half behind, and nothing in the number is a grade.

0 Quick Answer

An IQ of 100 is the mean of the norm sample by definition: it is the 50th percentile of the reference group, the exact centre of the Average band that runs from 90 to 109, and the one score on the scale that says a person performed like the middle of the crowd, no better and no worse. A modern IQ is a deviation score. A test is given to a reference sample, the mean raw performance of each age band is set to 100 and the spread to 15 points, and every later score is a distance from that centre. A person at 100 matched the middle of their age band exactly. About 68 percent of people score between 85 and 115, about 95 percent between 70 and 130, and about half between 90 and 110, by our arithmetic from the normal curve. The page on what an IQ is explains the scale as a whole; this page is about the number in the middle of it.

Four qualifications turn the definition into a reading. First, a 100 is an interval: the ACIS Full Scale IQ has a standard error of 1.60 points, so a reported 100 is about 97 to 103 at 95 percent confidence, our arithmetic, and a 98 cannot be told apart from a 102. Second, a 100 belongs to its norm sample. It equals a 100 on another test only if the two reference samples are equivalent, and a 100 earned against norms drawn in 1990 corresponds to about 92 against norms drawn in 2024, because raw performance rose by about 2.31 points per decade across the 285 studies in Trahan and colleagues' meta-analysis of the Flynn effect, our arithmetic on their figure. Third, 100 is not the world average: every population sits at 100 on norms drawn from itself, which is why the page on average IQ in the UK can say that the British mean is 100 by construction. Fourth, a 100 predicts average outcomes with wide scatter. In Strenze's meta-analysis of longitudinal studies intelligence correlates 0.56 with later education, 0.45 with occupation and 0.23 with income, and in Zagorsky's analysis of the NLSY79 the group scoring around 100 had a median income of 36,826 dollars against a sample median of 35,918.

What to do with a 100 is therefore simple: nothing with the number, which is the definition of typical, and everything with the profile of domain scores around it, since an average Full Scale score can sit on top of a verbal index of 114 and a processing speed index of 88, as the worked example below shows.

50th

The percentile of an IQ of 100 on any scale centred at 100: half of the reference sample scored above it and half below, whatever the standard deviation.

97 to 103

The 95 percent confidence interval around an ACIS Full Scale IQ of 100, from a standard error of 1.60 points on a composite reliability of .9886, our arithmetic.

49 percent

The share of a normed population inside the Average band of 90 to 109, treating each integer score as running half a point either side, our arithmetic from the normal curve.

92

What a 100 earned against norms drawn in 1990 corresponds to against norms drawn in 2024, at the 2.31 points per decade of the Flynn effect meta-analysis, our arithmetic.

1 What 100 Means: The Mean of the Norm Sample by Definition

A score of 100 carries no information about ability in the abstract; it marks the point at which a test's norm sample was centred, and everything the number says is said relative to that sample. The arithmetic is short. A raw score, the number of items answered correctly or the points earned for the quality of answers, is compared with the distribution of raw scores among people of the same age in the reference sample. The distance from the sample mean is expressed in standard deviation units, called a z score, and the IQ is 100 plus 15 times that z score. A z score of zero is an IQ of 100, and a z score of zero means the raw performance equalled the mean of the age band. The page on how IQ scores are normed describes how the tables are built; the point here is that 100 is where the builder put the origin.

Two things follow at once. The first is that the average IQ of the population a test was normed on is 100 at the time of norming and cannot be anything else, which is why the page on average IQ answers its own question by construction rather than by survey. David Wechsler, who introduced the deviation score in 1939, wrote in The Measurement of Adult Intelligence that the numerical value of an IQ is a matter of convention, that the score merely defines a person's relative position within the group they were compared with, and that the mean had been set by history "once and for all at 100". The second consequence is that a 100 refers to a particular sample. ACIS scores an adult against an English speaking reference frame of 3,243 records aged 16 to 90, described in the technical manual, with a technical analysis set of 2,750 complete records and a mean age of 33.91. A 100 on ACIS means the middle of that frame's age band, and a 100 on a Wechsler scale means the middle of the publisher's standardization sample at the same age. The two are the same statement about two different crowds, and they coincide only to the extent that the crowds do.

A third consequence is less obvious. In a normal distribution the mean, the median and the mode coincide, so 100 is not only the average score but the most common single score and the one that splits the population into equal halves. A score of 100 is therefore never rare and never one in anything. It is the reference point from which every other score is measured, and the distance from it, rather than the number itself, is what the rest of this page is about.

2 The Arithmetic of the Middle: 50th Percentile, 68 Percent, 95 Percent

Because the scale is a normal distribution with a mean of 100 and a standard deviation of 15, the population around 100 is fixed by arithmetic rather than by any test, and the same few numbers describe it on every instrument that uses the scale. About 68 percent of a normed population lies within one standard deviation of the mean, between 85 and 115. About 95 percent lies within two, between 70 and 130, and about 99.7 percent within three, between 55 and 145. The middle half of the population lies between 89.9 and 110.1, by our arithmetic, which is the origin of the round statement that half of all people score between about 90 and 110. The page on the standard deviation of 15 explains why that unit was chosen, and the page on the IQ score chart lays out every band on the site's public table.

The percentile of each score follows from the same curve. A 100 is the 50th percentile. A 90 is about the 25th and a 110 about the 75th, so the band between them holds the middle half. An 85 is the 16th and a 115 the 84th; a 120 is about the 91st, a 125 about the 95th, a 130 about the 98th and a 145 about the 99.9th, all by our arithmetic. The table below sets the bands of the ACIS public classification beside their percentiles, the share of the population each holds and the descriptor a Wechsler manual would attach to the same range. The shares treat each integer score as running half a point either side of itself, so that 90 to 109 covers 89.5 to 109.5, and they sum to 100 percent.

Score bandPercentile at the band's edgesShare of population, our arithmeticACIS public labelWechsler style descriptor, fourth edition wording
69 and belowBelow the 2ndAbout 2.1 percentMild, Severe and Profound Impairment (55 to 69, 40 to 54, below 40)Extremely Low
70 to 792nd to 8thAbout 6.5 percentBorderlineBorderline
80 to 899th to 23rdAbout 15.6 percentLow AverageLow Average
90 to 10925th to 73rdAbout 49.5 percentAverageAverage
110 to 11975th to 90thAbout 16.6 percentHigh AverageHigh Average
120 to 12991st to 97thAbout 7.2 percentSuperiorSuperior
130 to 13498th to 99thAbout 1.4 percentModerately GiftedVery Superior
135 to 14499.0th to 99.8thAbout 0.9 percentHighly GiftedVery Superior
145 to 15999.9th and aboveAbout 0.15 percentExceptionally GiftedVery Superior
160 to 177Above the 99.997thAbout 0.004 percentProfoundly Gifted (160 to 174 and 175 to 177)Very Superior

Read the table from the middle outward. The Average band alone holds about half of everyone, the two bands beside it hold another third between them, and the whole of the gifted range above 130 holds about one person in forty four. A 100 sits where the curve is highest, so small differences near it are common and mean little, while the same differences in the tails are rare and mean a great deal. The rarity calculator turns any score into one person in how many; at 100 the answer is one in two.

3 Where the Centre Came From: Stern's Quotient, Terman's 100, Wechsler's 90

The number 100 is a fossil of a formula: it began as 100 percent, the quotient of a child whose mental age equalled their chronological age, and it survived the abandonment of the quotient because Wechsler saw no reason to move it. The Binet-Simon scale, in its 1908 revision, expressed a child's performance as a mental age, the age at which the average child passed the same items. In 1912 William Stern proposed dividing that mental age by the chronological age, and the 1914 English translation of his Psychological Methods of Testing Intelligence defines the mental quotient as mental age divided by chronological age, with a value of 1 for a child at the normal level of their age, a value above 1 for an advanced child and a proper fraction for a child behind. An eight year old with a mental age of six had a quotient of 0.75.

Lewis Terman multiplied the fraction by 100 in his 1916 Stanford revision, The Measurement of Intelligence, and stated the design rule that made 100 the centre: a correct scale must give a median intelligence quotient of unity, or 100 percent, to unselected children of each age. The 100 was literally a percentage, and the median child of any age scored it by construction of the scale. Terman's classification put 90 to 110 under the label of normal or average intelligence, 110 to 120 under superior, 120 to 140 under very superior and above 140 under genius or near genius, and his distribution of 1,000 unselected children showed the lowest 1 percent at 70 or below, the highest 1 percent at 130 or above and the middle half between about 92 and 108. The page on the history of IQ follows the scale from Paris to Stanford, and the page on mental age explains why the quotient fails for adults, whose test performance stops rising in the late teens while their age does not.

Wechsler's 1939 deviation score dropped the quotient and kept the number. He had to choose an IQ for the mean of each age group and an IQ for some fixed distance below it, and he reasoned that the mean was settled by history at 100 while the lower anchor should follow existing usage, in which an IQ of 90 had come to be read as the bottom of average intelligence. He therefore equated 90 with one probable error below the mean, the lower boundary of the middle 50 percent. A probable error of 10 points is a standard deviation of 14.8, our arithmetic on the 0.6745 ratio between the two measures, and every Wechsler scale since has fixed the standard deviation at exactly 15. The two conventions that define the modern scale, 100 for the centre and 15 for the unit, are thus a percentage from 1916 and a boundary of average from 1939, and neither was ever a measurement of anything.

4 Why Average Is a Band From 90 to 109 and Not a Point

Average is a band rather than a point because no test can separate a 98 from a 102, and the band's edges were set by a convention about the middle half of the population that is older than the deviation IQ itself. Wechsler's 1939 classification defined each level by its distance from the mean in probable errors, and defined Average as the interval from one probable error below the mean to one above, the middle 50 percent of the population, which on the Wechsler Bellevue scale ran from 91 to 110. Terman had used 90 to 110. The Wechsler adult scales have since used 90 to 109 as the Average band, and the ACIS public classification adopts the same boundaries, with High Average from 110, Superior from 120 and the gifted bands from 130 upward, as the table above sets out. The guide to the 90 to 109 band covers the whole range, and the page on what IQ scores mean explains how to read any band with its uncertainty.

The band is wide because the middle of the curve is crowded. Twenty integer scores from 90 to 109 hold about 49 percent of the population by our arithmetic, and the ten scores from 110 to 119 hold about 17 percent, so a person at 100 has about twice as many people within ten points of them as the High Average and Superior bands contain together. A 100 is not at the edge of anything. It is ten points from the start of High Average and eleven from the top of Low Average, and with a standard error of 1.60 on the ACIS Full Scale score those edges are six to seven standard errors away, so a reported 100 is Average with near certainty even though the exact point is not certain at all.

The width also protects against the commonest misreading, which is to treat the label as a grade. Average on this scale is not a C. It is the modal condition of the population, the range the scale was built around, and the range for which Wechsler chose the word precisely because it described the middle half of everyone. The labels above and below it describe distance from that middle, not merit, and a report that prints Average beside a 100 is stating a tautology: the score is at the centre, and the centre is average by definition.

5 A 100 Is Really an Interval: The Standard Error

Every reported 100 is an estimate with a standard error, and the honest form of the number is an interval whose width depends on the reliability of the score that produced it. The standard error of measurement, as Harvill's instructional module for the National Council on Measurement in Education sets out, equals the standard deviation of the scale multiplied by the square root of one minus the reliability of the score. A band of one standard error either side of an observed score contains the true score about 68 percent of the time, and a band of 1.96 standard errors contains it about 95 percent of the time. Harvill's argument was that the standard error, not the reliability coefficient, is the statistic a reader of an individual score needs, because it is expressed in score points.

For the ACIS Full Scale IQ the composite reliability is .9886, so the standard error is 15 multiplied by the square root of 0.0114, which is 1.60 points, and the 95 percent interval around a reported 100 runs from about 97 to 103, our arithmetic. In percentile terms the true standing of a person reported at 100 very probably lies between about the 42nd and the 58th percentiles. Two people reported at 98 and 102 differ by four points; the standard error of the difference between two such scores is 1.60 multiplied by the square root of two, about 2.3 points, so the 95 percent margin is about 4.4 points and the two cannot be told apart on this evidence. The page on reliability and validity explains the coefficients, and the technical manual reports the error of every score.

Domain indices are wider. The ACIS Working Memory Index has a standard error of 4.12 points, so a reported 100 spans about 92 to 108 at 95 percent; the Processing Speed Index has a standard error of 5.22, so a 100 spans about 90 to 110; the Fluid Reasoning Index, at 2.48, and the Verbal Comprehension Index, at 2.40, each span about 95 to 105, all our arithmetic. An index of 100 can therefore have a true value in the Low Average or the High Average band, which is why the manual's own guidance is to state a range rather than a point. Measurement error is not the only source of movement. A second sitting of the same battery produces practice gains, conditions on the day move speeded scores by several points, and the norm tables themselves carry sampling error, so a 100 read a year apart on two tests could honestly appear as 96 and 104 without either sitting being wrong.

6 The Same 100 on Different Scales and Norms

A 100 is the mean on every scale that centres at 100, whatever its standard deviation, but one step away from the centre the scales disagree, and one change of norm year moves even the centre. A z score of zero is 100 on the Wechsler scales, on ACIS, on the older Stanford-Binet editions with their standard deviation of 16 and on the Cattell scale with its standard deviation of 24, so a 100 is the 50th percentile on all of them; one standard deviation up, at the 84th percentile, the four scales read 115, 115, 116 and 124. The page on the IQ percentile chart gives the full conversion, and the table sets the same standing side by side, using an ACIS 115 as the reference point one standard deviation up.

Scale or situationCentre and spreadWhat a 100 means thereWhat an ACIS 115 corresponds to
Wechsler scales and ACISMean 100, standard deviation 15The mean of the norm sample, 50th percentile115
Stanford-Binet, editions before the fifthMean 100, standard deviation 16The mean, 50th percentile116
Cattell scale used by some high IQ societiesMean 100, standard deviation 24The mean, 50th percentile124
Ratio IQ, Terman 1916Mental age divided by chronological age, times 100Mental age equal to chronological ageNo fixed conversion to a deviation scale
Subtest scaled scoreMean 10, standard deviation 31013
T scoreMean 50, standard deviation 105060
z scoreMean 0, standard deviation 101.0
Percentile rank1 to 9950th84th
A 100 earned against norms drawn in 1990, read in 2024Same scale, older sampleThe middle of the 1990 sample; about 92 against norms drawn in 2024, at 2.31 points per decade, our arithmeticAbout 107 against 2024 norms
A 100 on a test normed on a different populationMean 100 in that sampleThe middle of that sample; equal to an ACIS 100 only if the samples are equivalent115 in that sample's terms
A 100 from an unnormed online quizUndefinedNothing: no reference sample and no standard errorNot applicable

Mensa International's testing FAQ, read on September 19, 2026, makes the same point from the admissions side when it notes that a result of 132 on one test can be the same as a score of 148 on another, which is why the society states its threshold as the top two percent rather than as a number. Percentiles are the language for comparing standing across scales, and a 100 is the only score that needs no conversion, because the 50th percentile is 100 on every scale that centres there. The rest of the table is the reason to ask three questions of any other score: which scale, which norm sample, and which year.

7 100 on 1990 Norms Is Not 100 on 2024 Norms: The Flynn Effect

Norms date, and a 100 is stamped with the year its reference sample was drawn as surely as a price is stamped with a currency. James Flynn's 1987 paper in Psychological Bulletin, Massive IQ gains in 14 nations, assembled data from fourteen countries showing raw performance on intelligence tests rising by 5 to 25 points in a single generation, and the effect has carried his name since. Trahan, Stuebing, Fletcher and Hiscock's meta-analysis of 285 studies covering 14,031 people put the average gain at 2.31 points per decade, and at 2.93 points per decade on the modern comprehensive batteries, the Wechsler and Stanford-Binet scales, that most clinical scores come from. The mechanism for the number in the middle is simple. When a test is renormed, the new sample performs better than the old one in raw terms, the new mean is set to 100, and a raw performance that earned 100 against the old table earns less against the new one.

The arithmetic for a 100 is therefore a matter of dates. A person scoring 100 against norms drawn in 1990 is at the middle of the 1990 sample. Against norms drawn in 2024, 3.4 decades later, the same raw performance sits about 7.9 points lower at the meta-analytic rate, which is a score of about 92, and about 10 points lower at the Wechsler rate, a score of about 90, both our arithmetic on Trahan and colleagues' figures. The reverse holds too: a person at 100 on the 2024 table would have scored about 108 against the 1990 one. Neither score is wrong. Each is the middle of its own crowd, and the crowds performed differently. Pearson's WAIS-5, published in 2024 for ages 16 years 0 months to 90 years 11 months according to its product page read on September 19, 2026, is the latest resetting of the Wechsler centre, and a 100 on it is the middle of a sample drawn in the early 2020s.

The gains are not uniform, and the page on the Flynn effect reviews the evidence that they have slowed or reversed in some countries since the 1990s, which is why the correction above is an estimate rather than a rule. The practical consequence for a reader is a single question to put to any 100: in which year were the norms drawn? A score from a test normed twenty or thirty years ago flatters the person who takes it today by several points, a score from a freshly normed test does not, and two 100s from tests normed a generation apart describe two different standings within two different populations.

8 100 Is Not the World Average

There is no world norm sample, so there is no world average IQ, and the sentence that the average IQ is 100 is true of every normed population separately and of the planet not at all. The mean of any population on norms drawn from that population at that time is 100, in Delhi and in Leeds alike, because that is what norming does. The tables of national IQs that circulate are built differently: samples from many countries are scored against the norms of one, so that the numbers are distances from an anchor rather than the means of separate scales. In Becker's NIQ dataset, version 1.3.5, the United Kingdom sits at the anchor, at 99.66 on the combined measure and at exactly 100 for the conversion of school assessments, because the British standardization samples of Raven's matrices are the reference against which other countries' samples were placed. India is listed at 76.14 from convenience samples of children scored against British norms of other decades. The page on average IQ in India traces that figure to its samples, and the page on average IQ in the UK explains why the British figure could not have been anything but 100.

The site's own country ranking uses a different kind of table, the International IQ Test country data, in which the takers of one online test are grouped by country and the mean of the whole dataset is fixed at 100 by the site that runs it. That is a third meaning of 100 again: the middle of a self-selected group of people who chose to sit a particular test, not the middle of any nation. None of the three tables measures a world average, because no instrument has been normed on a probability sample of the world's adults, and no 100 on any of them describes a person.

The consequence for a reader with a 100 is a relief from a common anxiety. A 100 on a test normed in one country does not become a different number because the reader lives in another; it remains the middle of the sample the test was normed on, and the question to ask is which sample that was. A 100 on ACIS is the middle of an English speaking adult frame of 3,243 records, which is not a national sample of any country and does not claim to be. A 100 on a Wechsler scale with American norms is the middle of an American sample. The two agree to the extent that the samples perform alike, and the international tables have nothing to add to either.

9 What a 100 Predicts: School, Work and Income

A 100 predicts average outcomes in the only sense a correlation can: the expected education, occupation and income of people at 100 sit close to the population's own averages, with a spread around them nearly as wide as the population's. Strenze's 2007 meta-analysis of longitudinal studies in Intelligence put the correlation between intelligence measured early in life and later educational attainment at 0.56, later occupational level at 0.45 and later income at 0.23. Squared, those coefficients say that intelligence accounts for about 31 percent of the variance in education, 20 percent in occupational level and 5 percent in income, our arithmetic, leaving the rest to everything else. Deary, Strand, Smith and Fernandes' study of more than 70,000 English pupils found a latent correlation of 0.81 between general ability at eleven and examination results at sixteen, so a child at 100 at eleven is expected near the middle of the examination distribution five years later, with wide scatter around that expectation. The page on IQ and academic achievement examines the relationship.

For work, Sackett, Zhang, Berry and Lievens' reanalysis of the selection literature in the Journal of Applied Psychology in 2022 put the validity of general mental ability for job performance at about 0.31, down from the 0.51 that the field had quoted since 1998, and the page on IQ and job performance sets out why. Gottfredson's Why g matters used Wonderlic data from 1992 to show where a 100 sits among job applicants: the median applicant IQ was about 100 for cashier, meter reader and teller, about 90 for custodian and material handler, about 112 for accountant and teacher and about 120 for research analyst and advertising manager, and the middle half of applicants to any one job spanned 15 to 20 points, so every occupation overlaps its neighbours. The page on average IQ by profession gives the fuller tables.

For money the evidence is the most specific. Zagorsky's analysis of the NLSY79, 7,403 American baby boomers whose finances were measured in the 2004 survey round against an Armed Forces Qualification Test score rescaled to a mean of 100 and a standard deviation of 15, found that each IQ point was associated with 234 to 616 dollars more income per year after controls, that IQ correlated 0.30 with income and 0.16 with net worth, and that after controls there was no statistically distinguishable relationship between IQ and wealth at all. The group scoring around 100 had a median income of 36,826 dollars and a median net worth of 57,550 dollars against sample medians of 35,918 and 55,250: the middle of the distribution, in money. The difference between a person at 100 and one at 130 was 6,000 to 18,500 dollars a year, and financial distress, in the form of missed payments, maxed cards and bankruptcy, was related to IQ in a curve rather than a line, with higher scores sometimes raising the probability of difficulty. The page on IQ and income reviews why the money correlation is the weakest of the set.

10 What a 100 Predicts: Everyday Life and Health

The clearest published statement of what an average score means in daily life is Gottfredson's, that the middle half of the distribution is readily trained for the bulk of jobs in society, and its life outcomes sit close to the population's own. Her 1997 paper reproduced Herrnstein and Murray's tabulation of NLSY79 outcomes for young White adults by IQ band, and the 91 to 110 band reads as follows: 6 percent had dropped out of high school, against 35 percent in the 76 to 90 band and 0.4 percent in the 111 to 125 band; 6 percent lived in poverty, against 16 percent below and 3 percent above; 8 percent of mothers were chronic welfare recipients, against 17 and 2; and 81 percent were married by thirty. The rates in the middle band were several times those of the band above and a fraction of those of the band below, which is the shape a moderate correlation produces: average risk, neither low nor high, with wide individual variation inside every band. Gottfredson's own caution was that these are odds rather than guarantees, and that individuals beat or fall behind the odds for their IQ level on other traits.

On trainability, the same paper placed people at roughly 100 to 113 on the Wonderlic conversion in the group able to learn routines quickly from a combination of written materials and job experience, and people below about 95 in the group that tends to need explicit teaching of most of what they must know. Her argument for why g matters was about the complexity of everyday tasks, from reading a schedule to following a physician's instructions, and her point about the middle band was that it handles most of them and can be trained for most jobs, which is a description of the modal adult and nothing more.

On health the evidence is a whole birth cohort. Calvin and colleagues followed 65,765 Scots born in 1936 and tested at eleven in 1947 through 68 years of death records, and reported in the BMJ in 2017 that each standard deviation of higher childhood intelligence was associated with a hazard ratio of 0.72 for death from respiratory disease, 0.75 for coronary heart disease, 0.76 for stroke and 0.84 for dementia, with small attenuation after adjustment for childhood socioeconomic status. On those ratios a person at 100 had, on average, a hazard of respiratory death 28 percent lower than a person at 85 and about 39 percent higher than a person at 115, our arithmetic on the published ratio, and the page on IQ and longevity reviews the field that grew from the cohort. These are population associations from childhood scores, and they say nothing about the lifespan of any one person at 100, whose health is determined by the things the correlation leaves out.

11 An Average Full Scale Can Hide an Uneven Profile

A Full Scale IQ of 100 is a summary, and two people who share it can have profiles twenty points apart in opposite directions, which is why the domain indices, not the headline, are the informative part of a report. ACIS builds the Full Scale score from 20 subtests across six domains, verbal comprehension, fluid reasoning, quantitative reasoning, visual spatial ability, working memory and processing speed, each reported as an index on the same scale of 100 and 15, as the page on the six cognitive domains describes. The Full Scale score is the best single estimate of the general factor that runs through all of them, with a loading of .958 in the ACIS confirmatory factor analysis, and the page on Full Scale IQ explains what it does and does not summarize. It does not summarize the shape of the profile.

Illustrative exampleThe report below is constructed to make a point and does not describe any person. A Full Scale IQ of 100 with an interval of 97 to 103; a Verbal Comprehension Index of 114, interval 109 to 119; a Fluid Reasoning Index of 102; a Quantitative Reasoning Index of 98; a Visual Spatial Index of 101; a Working Memory Index of 96; and a Processing Speed Index of 88, interval 78 to 98. All intervals are 95 percent bands from the ACIS standard errors, our arithmetic.

Read the profile before the total. The verbal index of 114 sits at about the 82nd percentile, and the processing speed index of 88 at about the 21st. The difference between them is 26 points. The standard error of a difference between two indices is the square root of the sum of their squared standard errors, which for the verbal index at 2.40 and the speed index at 5.22 is about 5.7 points, so a difference of about 11 points would be needed for significance at 95 percent, and a difference of 26 is more than four times the error, our arithmetic. This is a person whose acquired knowledge and verbal reasoning run comfortably above the middle and whose speed on simple timed decisions runs well below it, and the Full Scale 100 is the average of a strength and a weakness that describes neither. The manual's guidance for a lower speed index is to read it beside the other indices and beside the conditions of the sitting, because speed tasks are the most sensitive to devices, interruptions and fatigue.

Now set the same total on a flat profile: 101, 99, 100, 102, 98 and 100 across the six indices, no difference beyond error, a person for whom every kind of task is about equally easy. The two reports print the same number in the largest type and describe different people, and only for the second is the Full Scale score the whole story. A 100 is a sum. The profile says what was added.

12 Why a 100 From an Unnormed Quiz Is Not a Score

A 100 means the middle of a norm sample, so a number from a quiz that has no norm sample means nothing, however carefully it is printed. A score on the scale this page describes requires four things: a reference sample that is described, with its size, ages and language; norm tables that convert raw performance to a position within that sample by age band; a reported reliability and standard error for each score; and a stated centre and spread. A quiz that prints 100 without them has either fixed the mean of its own visitors at 100, which makes the number the middle of whoever happened to take it, or applied a formula to a raw count, which makes it the middle of nothing. The page on free versus validated IQ tests lists what to look for, and the page on whether online IQ tests are accurate explains what a credible online battery publishes.

The high IQ societies apply the same test. Mensa International's testing FAQ, read on September 19, 2026, states that membership requires a score in the top two percent of the general population on an approved intelligence test that was properly administered and supervised, and that online tests cannot be used for admission. The page on the Mensa IQ test sets out the route. The principle that makes an unnormed 100 meaningless makes an unnormed 130 meaningless too, and the societies simply refuse both.

We sell a competing paid assessment, so treat this paragraph as a disclosure and check it against the rest of the page. ACIS is an online, self administered battery of 20 subtests. It publishes its reference frame, 3,243 English speaking adult records aged 16 to 90 with a technical analysis set of 2,750, its reliability coefficients, the standard error of every index and a confirmatory factor analysis, in a technical manual, version 1.4, and it prints the interval beside every score in a report. It is not a national probability sample of any country and does not claim to be, its processing speed index is the score most sensitive to device and conditions, and its norms describe the frame they were built on and no other. A 100 on it is therefore a number with a definition, an error and a sample attached, which is the whole difference between a score and a quiz result.

13 What to Do With a 100

The correct response to a Full Scale IQ of 100 is to do nothing with the number and everything with the profile, because 100 is the definition of typical and carries no instruction. It is not a diagnosis, a ceiling or a verdict. It does not qualify for Mensa, which requires the 98th percentile, a 130 on this scale, and it does not disqualify from anything, since the applicant medians for most jobs sit within fifteen points of it. It does not fix the future, because the correlations of the previous sections leave most of the variance in every outcome to other things. And it is not permanent. Ritchie and Tucker-Drob's meta-analysis of quasi-experimental studies found that each additional year of education raises measured intelligence by about one to five points, the page on whether you can improve your IQ reviews the short list of things that move a score, and the page on whether IQ changes with age explains why the standing behind a 100 is fairly stable across adulthood without being fixed.

What is useful is the profile. A report that prints six indices around a 100 says which kinds of task were easy and which took effort, and where the indices differ by more than their combined error the difference is a real feature of the person and not of the test. That is the information a reader can act on: a verbal index well above the middle points to the tasks that will feel natural, a speed index below it points to the conditions worth controlling, and a flat profile says that the total is the whole story. The number in the middle says only that the sum came out average.

If the 100 came from a quiz without norms, the useful step is to take a battery with them and read the result with its interval. ACIS offers three forms at one time prices read on September 19, 2026: the Full Scale form of 20 subtests at 50 dollars, the Optimized form of 13 subtests at 30 dollars and the Quick form of 6 subtests at 15 dollars, with a free trial of five subtests that asks for no card, 30 days to complete a form and a 5 day quality guarantee. The prices can change. Whatever instrument a reader uses, the reading of a 100 does not change: it is the middle of the sample the test was normed on, an interval of a few points either side, a standing that other tests and other decades would express differently, and the starting point for the profile, which is where the person is.

14 Sources Behind This Page

Every figure above is traceable to one of the following, and each is linked at the point where it is used. ACIS reliability, standard error and loading figures are from the technical manual, version 1.4; every percentile, population share, confidence interval, Flynn correction and squared correlation is our arithmetic on a published figure and is labelled as such where it appears. Product details and society rules were read on the pages cited on September 19, 2026 and will change.

  • Stern W. The Psychological Methods of Testing Intelligence. Warwick and York, 1914, translated by Guy Montrose Whipple. archive.org.
  • Terman L M. The Measurement of Intelligence: An Explanation of and a Complete Guide for the Use of the Stanford Revision and Extension of the Binet-Simon Intelligence Scale. Houghton Mifflin, 1916. archive.org.
  • Wechsler D. The Measurement of Adult Intelligence. Williams and Wilkins, 1939. archive.org.
  • Flynn J R. Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin, 1987, volume 101, issue 2, pages 171 to 191.
  • Trahan L H, Stuebing K K, Fletcher J M and Hiscock M. The Flynn effect: A meta-analysis. Psychological Bulletin, 2014, volume 140, issue 5, pages 1332 to 1360.
  • Harvill L M. Standard error of measurement. Educational Measurement: Issues and Practice, 1991, volume 10, issue 2, pages 33 to 41.
  • Gottfredson L S. Why g matters: The complexity of everyday life. Intelligence, 1997, volume 24, issue 1, pages 79 to 132.
  • Strenze T. Intelligence and socioeconomic success: A meta-analytic review of longitudinal research. Intelligence, 2007, volume 35, issue 5, pages 401 to 426.
  • Zagorsky J L. Do you have to be smart to be rich? The impact of IQ on wealth, income and financial distress. Intelligence, 2007, volume 35, issue 5, pages 489 to 501.
  • Deary I J, Strand S, Smith P and Fernandes C. Intelligence and educational achievement. Intelligence, 2007, volume 35, issue 1, pages 13 to 21.
  • Calvin C M, Batty G D, Der G, Brett C E, Taylor A, Pattie A, Čukić I and Deary I J. Childhood intelligence in relation to major causes of death in 68 year follow-up: prospective population study. BMJ, 2017, volume 357, article j2708.
  • Ritchie S J and Tucker-Drob E M. How much does education improve intelligence? A meta-analysis. Psychological Science, 2018, volume 29, issue 8, pages 1358 to 1369.
  • Sackett P R, Zhang C, Berry C M and Lievens F. Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range. Journal of Applied Psychology, 2022, volume 107, issue 11, pages 2040 to 2068.
  • Becker D. The NIQ dataset, version 1.3.5, released October 25, 2023. viewoniq.org, read September 19, 2026.
  • Pearson. Wechsler Adult Intelligence Scale, Fifth Edition (WAIS-5), product page. pearsonassessments.com, read September 19, 2026.
  • Mensa International. Getting your IQ tested: frequently asked questions. mensa.org, read September 19, 2026.

15 Frequently Asked Questions

Is an IQ of 100 good?

It is average by definition, which is neither good nor bad. A 100 is the mean of the norm sample and the 50th percentile: half of the reference group scored above it and half below. On the ACIS public bands it sits in the exact centre of Average, which runs from 90 to 109 and holds about half of everyone.

What percentile is an IQ of 100?

The 50th. On any scale centred at 100 a score of 100 has a z score of zero, so exactly half of the reference population lies above it and half below, whatever the standard deviation. A 90 is about the 25th percentile and a 110 about the 75th, so the middle half of people lies between those two scores.

What does an IQ of 100 mean?

It means the person's raw performance matched the mean of their age band in the test's norm sample. An IQ is a deviation score, 100 plus 15 times the distance from that mean in standard deviation units, so a 100 is a distance of zero. The number describes a position within a reference group, not a quantity of intelligence.

Is 100 the average IQ?

Yes, by construction. When a test is normed, the mean of the reference sample is set to 100 and the standard deviation to 15, so the average member of the population the test describes scores 100 at the time of norming. The figure carries no information about the population; it is the origin of the scale.

Is 100 the world average IQ?

No, because no test has been normed on the world. Every population averages 100 on norms drawn from itself, and the national tables that circulate score samples from many countries against the norms of one, with Britain as the anchor in Becker's dataset. A 100 is the middle of one specific norm sample and of nothing larger.

Is an IQ of 100 high average or low average?

Neither. High Average is 110 to 119 and Low Average is 80 to 89 on the Wechsler style bands that ACIS also uses, and 100 lies in the middle of the Average band between them, ten points from the start of High Average and eleven from the top of Low Average. It is the centre of the centre.

How common is an IQ of 100?

It is the most common single score. In a normal distribution the mean, the median and the mode coincide, so more people score 100 than any other value, about 49 percent of the population falls in the 90 to 109 band, and about 68 percent falls between 85 and 115, by our arithmetic from the curve.

Why is the average IQ 100 and not 50 or 1,000?

Because Lewis Terman multiplied William Stern's mental quotient by 100 in 1916, so that a child whose mental age equalled their chronological age scored 100 percent. When David Wechsler replaced the quotient with a deviation score in 1939 he kept the historical centre of 100 and anchored 90 as the bottom of average.

Why is the Average band 90 to 109?

Because Wechsler defined Average in 1939 as the middle 50 percent of the population, one probable error either side of the mean, which ran from 91 to 110 on his scale, and the later Wechsler adult scales settled on 90 to 109. Terman had already used 90 to 110 in 1916. About 49 percent of people fall inside the band.

Is a 100 on one IQ test the same as a 100 on another?

Only if the two norm samples are equivalent. A 100 is the mean of a specific reference sample, so a 100 against American Wechsler norms, a 100 against the ACIS adult frame and a 100 on a test normed decades ago each mark the middle of a different group. Comparing them requires knowing the sample and the year.

What is a 100 on 1990 norms worth today?

About 92 against norms drawn in 2024, our arithmetic from the 2.31 points per decade in Trahan and colleagues' meta-analysis of the Flynn effect, or about 90 at the 2.93 points per decade found on Wechsler and Stanford-Binet scales. Raw performance rose between the two samples, so the same performance sits lower against the newer table.

What income does an IQ of 100 predict?

About the median. In Zagorsky's analysis of the NLSY79 the group scoring around 100 had a median income of 36,826 dollars in 2004 against a sample median of 35,918, each IQ point was associated with 234 to 616 dollars more income per year, and IQ had no statistically distinguishable relationship with wealth after controls.

What jobs can someone with an IQ of 100 do?

Most of them. Gottfredson's analysis of Wonderlic applicant data put the median applicant IQ at about 100 for cashier, meter reader and teller, about 112 for accountant and teacher and about 120 for research analyst, with the middle half of applicants to any job spanning 15 to 20 points, so the occupations overlap widely.

Does an IQ of 100 predict health or lifespan?

Only as a population average. In 65,765 Scots tested at eleven in 1947 and followed for 68 years, each standard deviation of higher childhood intelligence was associated with hazard ratios of 0.72 for respiratory death, 0.75 for coronary heart disease, 0.76 for stroke and 0.84 for dementia. A person at 100 sits at the cohort's middle risk.

How accurate is a score of 100?

It is an interval. The ACIS Full Scale IQ has a standard error of 1.60 points, so a reported 100 spans about 97 to 103 at 95 percent confidence, our arithmetic, and a domain index with a standard error of 4 to 5 points spans about 90 to 110. Two scores of 98 and 102 cannot be told apart.

Can an IQ of 100 get into Mensa?

No. Mensa requires a score in the top two percent of the general population on an approved supervised test, which is about 130 on a scale with a standard deviation of 15 and 148 on the Cattell scale with a standard deviation of 24. A 100 is the 50th percentile, and online tests are not accepted in any case.

Can two people with an IQ of 100 be very different?

Yes. The Full Scale score is a sum across domains, so a person with a verbal index of 114 and a speed index of 88 can share a 100 with someone whose six indices all sit near 100. The 26 point gap in the first profile is more than four times its standard error and is real.

Can an IQ of 100 go up?

Measured scores move. Each additional year of education adds about one to five points in the quasi-experimental meta-analysis, raw performance in whole populations rose about 2.31 points per decade through the twentieth century, and practice and conditions on the day shift a single sitting by several points. Relative standing in adulthood is fairly stable but not fixed.

Is a 100 from a free online quiz a real IQ?

Not unless the quiz publishes a norm sample, reliability and a standard error and reports on a stated scale. A quiz that fixes the mean of its visitors at 100 reports the middle of whoever took it, and one that applies a formula to a raw count reports the middle of nothing. Neither is a position in a reference group.

What should I do with an IQ score of 100?

Nothing with the number and everything with the profile. A 100 is the definition of typical: it is not a diagnosis, a ceiling or a qualification for anything. The domain indices around it say which kinds of task were easy and which took effort, and differences between them that exceed their combined error are real information about the person.

What does a 100 on the ACIS test mean?

It means the person matched the middle of an English speaking adult reference frame of 3,243 records aged 16 to 90, with a technical analysis set of 2,750, on a Full Scale score built from 20 subtests with a reliability of .9886. The report prints the interval, about 97 to 103, and six domain indices with their own errors.

Take the assessment

You get a profile, not a number

ACIS measures six CHC domains across 20 subtests and reports each one with its own normed score and confidence interval, so you can see where you are strong and where you are not.

Free trial, no card required. Full report from $15.