Global IQ Research — Updated May 2026
IQ score by country data draws heavy traffic and heated debate. The most-cited dataset — the Becker-Lynn 2019 compilation of national IQ estimates — places East Asian nations at the top and shows a wide spread between high- and low-income regions. But the numbers are estimates built from heterogeneous studies, and the gap between a rigorous national sample and a single school study from 1985 is enormous.
This guide presents the ranked data, explains the environmental drivers behind national differences, covers the Flynn Effect and its recent reversals, and gives you a framework for reading any country-level IQ claim responsibly.
106.5
Top estimate
Japan & Taiwan, Becker-Lynn 2019
~3 pts
Per decade
Flynn Effect historical gain rate (20th century)
13.5
IQ points
Avg deficit from severe iodine deficiency
±15
Within-country SD
Individual variation inside any country
A national IQ estimate is not a direct measurement of every person in a country. It is an aggregated figure — typically a weighted average of one or more cognitive studies conducted in that country — rescaled onto the standard IQ metric where 100 represents the normed global midpoint and the standard deviation is 15.
The most widely used academic database is the Becker-Lynn National IQ dataset (viewoniq.org), which covers estimates for over 200 countries and was updated most recently in 2019. A parallel resource, Worlddata.info, blends cognitive study data with PISA international school assessment scores and produces somewhat different rankings — illustrating that methodology choice materially affects the output.
Country IQ tables combine data from clinical cognitive tests, school-age reasoning tests, military conscript assessments, PISA academic assessments, and in some cases single small studies conducted decades ago. These are not equivalent instruments. Understanding the source for any given country estimate is more informative than the number itself.
| Source type | Reliability | Key caveat |
|---|---|---|
| Nationally representative cognitive study (recent) | Highest | Still affected by language, norming year, test content |
| PISA international school assessments | High (achievement proxy) | Measures academic learning, not IQ — useful context |
| Military conscript reasoning tests | Moderate | Male-only; military-selection bias; often 1950s-1990s |
| Single regional or school sample | Low | One city or school type; not nationally representative |
| Online self-selected samples | Not reliable for countries | Self-selection and internet access bias invalidate comparisons |
High-income countries typically have multiple modern, well-funded cognitive studies with large nationally representative samples. Many lower-income countries have one study, conducted decades ago, on a convenience sample. The standard error around a single regional study can easily be 10 points or more — meaning two countries that look different by 8 points may not be statistically distinguishable at all.
Within any country, IQ scores follow a normal distribution with a standard deviation of approximately 15 points. That means a country with a national average of 100 contains many people scoring 130+ and many scoring below 70. The individual spread is massive. Two countries with averages of 97 and 103 will have enormously overlapping score distributions — the gap at the average tells you almost nothing about any specific person from either country.
Use country-level estimates for broad research questions about education, health, and economic development — not for judging individuals, supporting stereotypes, or making hiring or immigration decisions.
The table below draws on the Becker-Lynn 2019 National IQ dataset, the most comprehensive academic compilation available. PISA 2022 mathematics scores (released December 2023) are included as a parallel independent measure of academic reasoning performance. Scores should be read as estimates with uncertainty ranges, not precise measurements.
Source: Becker-Lynn 2019 National IQ dataset (viewoniq.org) • PISA 2022 (OECD, December 2023). China’s PISA participation is limited to selected high-performing provinces and is not nationally representative.
| # | Country | Region | Est. IQ |
|---|---|---|---|
| 1 | 🇯🇵Japan | East Asia | 106.5 |
| 2 | 🇹🇼Taiwan | East Asia | 106.5 |
| 3 | 🇸🇬Singapore | SE Asia | 105.9 |
| 4 | 🇭🇰Hong Kong | East Asia | 105.3 |
| 5 | 🇨🇳China | East Asia | 104.1 |
| 6 | 🇰🇷South Korea | East Asia | 102.3 |
| 7 | 🇫🇮Finland | Europe | 101.2 |
| 8 | 🇩🇪Germany | Europe | 100.7 |
| 9 | 🇳🇱Netherlands | Europe | 100.4 |
| 10 | 🇨🇭Switzerland | Europe | 100.0 |
| 11 | 🇦🇹Austria | Europe | 100.0 |
| 12 | 🇮🇹Italy | Europe | 99.9 |
| 13 | 🇧🇪Belgium | Europe | 99.5 |
| 14 | 🇸🇪Sweden | Europe | 99.3 |
| 15 | 🇨🇦Canada | N. America | 99.0 |
| 16 | 🇭🇺Hungary | Europe | 98.9 |
| 17 | 🇪🇪Estonia | Europe | 98.9 |
| 18 | 🇬🇧United Kingdom | Europe | 98.5 |
| 19 | 🇨🇿Czech Rep. | Europe | 98.2 |
| 20 | 🇦🇺Australia | Oceania | 98.0 |
| 21 | 🇫🇷France | Europe | 98.0 |
| 22 | 🇮🇪Ireland | Europe | 97.9 |
| 23 | 🇳🇿New Zealand | Oceania | 97.5 |
| 24 | 🇺🇸United States | N. America | 97.4 |
| 25 | 🇵🇱Poland | Europe | 97.0 |
† China participates in PISA only through selected provinces (Shanghai, Beijing, Jiangsu, Zhejiang) and results are not nationally representative. OECD average PISA 2022 Math: 472.
The clustering of East Asian countries at the top of both the IQ estimates and PISA rankings is consistent across most datasets and is well-replicated. Researchers point to high rates of educational investment, cultural emphasis on academic performance, strong test familiarity from early childhood, and well-funded school systems as partial explanations — alongside possible genuine population-level cognitive differences that remain contested and difficult to disentangle from environment.
Estonia is a notable outlier: it scores remarkably high on PISA (510 in Math, 511 in Reading — among the best in Europe) despite being a small country with a GDP per capita well below Switzerland or the Netherlands. This underscores that policy and educational system design, not just wealth, drive measured performance.
The United States ranks 24th in the Becker-Lynn IQ estimates (97.4) but shows a strikingly split PISA profile: 465 in Math (below the OECD average of 472) but 504 in Reading (above average) and 499 in Science. This pattern is consistent with structural features of the US education system: strong literacy emphasis, high inequality in school quality across states, and less systematic math curriculum standardization compared to high-performing Asian and European systems.
Estimated IQ from Becker-Lynn 2019. Bar width is scaled relative to the range 90–108. PISA 2022 Math score shown alongside.
IQ scale shown starts at 90 for visual clarity. All estimates carry uncertainty ranges not shown here.
The dominant scientific interpretation of national IQ differences is environmental, not genetic. The factors below are the most empirically supported, each with quantified effect sizes from peer-reviewed research. Click any factor to expand the evidence.
A 2018 meta-analysis in Psychological Science found that each additional year of formal education raises IQ scores by approximately 1 to 3 points. Two years of schooling showed larger cognitive gains than differences attributable to socioeconomic status. Countries that expanded secondary and tertiary education during the 20th century saw corresponding gains in measured test performance — a key driver of the Flynn Effect.
Key source: Ritchie & Tucker-Drob, Psychological Science, 2018
A meta-analysis of 18 studies found that populations with severe iodine deficiency score an average of 13.5 IQ points lower than non-deficient populations. Salt iodization programmes have achieved measurable population-level IQ gains: in iodine-deficient US regions, iodization was associated with roughly a 15-point increase in affected communities. This is one of the clearest demonstrations that measured national IQ differences reflect environmental inputs, not fixed traits.
Key source: Qian et al.; Feyrer et al. 2017 (AEJ: Applied Economics)
A 2022 study estimated that childhood lead exposure reduced the average IQ of over half of all living Americans by approximately 3 points, with the highest-exposed cohorts losing more than 7 points on average. Globally, children under 5 lost an estimated 765 million IQ points collectively in 2019 due to lead exposure (The Lancet Planetary Health, 2023). Countries that phased out leaded petrol earlier show corresponding cognitive-test improvements in subsequent cohorts.
Key source: Lanphear et al.; Attina & Trasande; McFarland et al. 2022 (PNAS)
Early-childhood nutrition — particularly protein, micronutrients (iron, zinc, iodine), and caloric sufficiency — has measurable effects on brain development and later cognitive test performance. The World Bank Human Capital Index explicitly incorporates stunting rates and child survival as predictors of adult productivity, partly because nutritional deficits in the first 1,000 days of life create lasting effects on measured intelligence.
Key source: World Bank Human Capital Project (2024 update)
Countries where standardized testing is deeply embedded in education culture — particularly South Korea, Japan, and Singapore — develop test-taking fluency that boosts performance on matrix-style IQ tasks. Translation quality, abstract reasoning exposure in schools, and whether children routinely practice timed multiple-choice tasks can shift scores by several points without reflecting any difference in underlying cognitive capacity.
Key source: Wicherts & Wilhelm, 2007; PISA methodology notes
Many national estimates in the most-cited databases draw on a single study, a regional sample, or data collected decades ago. Wicherts et al. (2010) demonstrated that selectively excluding available African studies produced estimates of ~70, while including all data raised the estimated average to ~82. A 12-point swing from data selection alone illustrates why estimates for data-sparse countries deserve extreme skepticism.
Key source: Wicherts, Dolan & van der Maas, Psych. Science in the Public Interest, 2010
When researchers statistically control for years of schooling, health, and nutritional status across countries, much of the variance in national IQ estimates is accounted for by these measurable inputs. This is not an argument that all differences are explained — the data is noisy enough that firm conclusions are difficult — but it strongly suggests that observed differences reflect unequal opportunity more than anything innate.
1–3 pts
Per additional year of schooling
Meta-analysis across 42 studies, Ritchie & Tucker-Drob, 2018
+15 pts
From iodine programme in deficient US regions
Feyrer et al., American Economic Journal: Applied, 2017
−7 pts
Avg loss in highest lead-exposed US cohorts
McFarland et al., PNAS, 2022
The World Bank Human Capital Index (2024 update) treats education, nutrition, health, and social protection as jointly determining adult cognitive and economic productivity. Countries that invest in all four pillars simultaneously tend to produce the largest gains in measured test performance — consistent with the idea that no single intervention alone drives national cognitive outcomes.
A country’s position in a ranking is a snapshot of its current measurable conditions — including its data quality — not a permanent property. Countries that rapidly expanded education access have shown rapid IQ gains within a single generation.
James Flynn documented one of the most important findings in 20th-century psychology: raw IQ test scores rose by approximately 3 points per decade across most countries for which data existed. A person of average intelligence in 1950 would score well below average against 2000 norms. This gain was not genetic — it happened too fast. It reflects the cumulative effects of better nutrition, more schooling, richer cognitive environments, reduced disease burden, and increasing test familiarity.
Reference point — baseline norms set
1940s
+6 pts vs 1940s across OECD nations
1960s
+12 pts — nutrition and schooling gains
1980s
+15 pts — levelling off in high-income countries
2000s
Reversal in Norway, Denmark, UK; gains continue in developing world
2020s
Illustrative only — bar heights represent approximate relative gain magnitude, not precise IQ values.
Since the 1990s, several high-income countries have reported stagnant or slightly declining IQ trajectories. Norway, Denmark, Finland, and the UK have all published data showing cohorts born after the mid-1970s scoring below the trend line established by earlier generations. Proposed explanations include changes in media consumption, reduced emphasis on deep reading, educational curriculum shifts, and possible dysgenic effects, though no consensus explanation has emerged.
Importantly, these reversals are not universal. Countries in sub-Saharan Africa, South and Southeast Asia, and parts of Latin America that have rapidly expanded educational access continue to show positive Flynn Effect trends. The pattern suggests the gain mechanism is still operating wherever its root causes — better nutrition, schooling, and test familiarity — are still improving.
If a country’s IQ estimate was measured in 1990 but its neighbour’s was measured in 2015, the comparison is partly confounded by 25 years of Flynn-Effect change. Comparisons across different measurement decades are especially unreliable — yet this is exactly what most country-ranking tables present without disclosure.
PISA (Programme for International Student Assessment), run by the OECD, tests 15-year-olds across 81 countries on mathematics, reading, and science every three years. While PISA measures academic achievement rather than IQ, its mathematics scale in particular correlates strongly with fluid reasoning performance and provides a more methodologically consistent cross-national benchmark than most cognitive datasets.
Source: OECD PISA 2022 Results Volume I (December 2023)
| Country | Math |
|---|---|
| Singapore | 575 |
| Japan | 536 |
| Taiwan | 547 |
| South Korea | 527 |
| Estonia | 510 |
| Switzerland | 508 |
| Canada | 497 |
| Netherlands | 493 |
| OECD Average | 472 |
| United States | 465 |
Estonia scored 510 in Math, 511 in Reading, and 526 in Science in PISA 2022 — consistently among the top European nations despite a GDP per capita roughly half that of Switzerland. Estonia reformed its entire school system after regaining independence in 1991, investing heavily in teacher quality, digital learning infrastructure, and curriculum coherence. Its PISA performance is one of the clearest demonstrations that system design, not wealth alone, drives cognitive outcomes.
Singapore scored 575 in Math — the highest in the world — alongside 543 in Reading and 561 in Science in PISA 2022. Its model combines structured curriculum progressions, highly selective teacher recruitment, strong parental investment in education, and systematic mastery-learning approaches. Singapore also tops the Becker-Lynn IQ estimates among Southeast Asian nations (105.9). The consistency across both measures makes it the world’s most reliably documented high-performance educational system.
The Lynn and Becker national IQ datasets are controversial within academic psychology. The central critiques, led by researchers including Jelte Wicherts (Tilburg University) and colleagues, are not about whether differences in measured test performance exist — they do — but about whether the published estimates accurately measure them.
The sampling critique
Wicherts, Dolan & van der Maas (2010, Psychological Science in the Public Interest) showed that Lynn’s sub-Saharan Africa IQ estimates (~70) depended on excluding the majority of available data. When all 37,000+ African test-takers in the literature were included, the average rose to approximately 82 — a 12-point revision driven entirely by data selection.
The methodology critique
National estimates mix tests from different years, languages, age groups, and psychometric traditions onto a single IQ scale. A 2022 research paper concluded that national IQ datasets “do not provide accurate, unbiased or comparable measures of cognitive ability worldwide.” Standard errors and mean IQ are negatively correlated (r ≈ −0.60), meaning the least reliable estimates tend to cluster at the bottom of rankings.
Mainstream cognitive scientists agree that measurable differences in average test performance exist across countries, and that environmental factors (education, nutrition, health) are the most empirically supported explanations. The scientific consensus does not support using national IQ estimates to make claims about innate group differences in intelligence, for the following reasons:
Check sample size, year, representativeness, and test type for each country estimate before interpreting it.
Prefer ranges or confidence intervals over point estimates — most country figures carry uncertainty of ±5-10 points.
Always consider education, health, and measurement quality as confounds before drawing comparative conclusions.
Never use country averages to make claims about individuals from those countries.
Treat estimates for data-sparse countries (fewer than 2-3 modern representative studies) as very weak evidence.
Where do you personally sit?
National averages are research estimates with wide uncertainty bands. Your individual score is determined by your own cognitive history, not your passport. The only way to know your actual reasoning baseline is to measure it directly — under controlled conditions, with a well-structured instrument.
IQMog uses a fixed Raven-style matrix reasoning dataset to give you a consistent, repeatable results report with score and percentile context after completion. No email required to start.
The most important statistical fact for interpreting country IQ tables: within any country, the standard deviation is approximately 15 IQ points. This means that in a country with an average of 100, about 16% of people score above 115 and about 16% score below 85. The full range from roughly 70 to 130 contains the vast majority of the population.
Score distribution overlap: Country A (avg 97) vs Country B (avg 106)
The two curves overlap almost entirely. Most people in the higher-average country score within the same range as most people in the lower-average country.
The gap between the United States (estimated 97.4) and Japan (estimated 106.5) is about 9 IQ points. In population terms, that shifts the average from the 47th percentile to the roughly 63rd percentile on a US-normed scale. But the score distributions of the two countries overlap by more than 75%. Millions of Americans score above the Japanese average; millions of Japanese score below the American average. A 9-point national difference is research-interesting but individually meaningless.
Your country’s estimated average is not your score. It is not a ceiling on your potential, a floor on your baseline, or a predictor of your performance. Individual reasoning ability is shaped by your specific education, health history, cognitive habits, sleep, and testing conditions — factors that vary enormously within every population. For a deeper look at what your score means in context, read the average IQ score guide or the full IQ score ranges breakdown.
Expand each question to reveal the evidence-based answer.
False — for multiple reasons.
First, within-country standard deviation (~15 points) dwarfs the between-country gap: most individuals in lower-ranked countries score above most individuals in higher-ranked ones. Second, the estimate for the top-ranked country carries uncertainty that may overlap with several countries below it. Third, “smartest” is not a meaningful summary of a population with a distribution spanning 70+ IQ points.
Not supported by evidence.
The Flynn Effect — 3 IQ points per decade over half a century within the same populations — proves that measured IQ responds strongly to environmental change. Iodization, lead removal, and educational expansion have all produced documented IQ gains. Genetic changes at the population level cannot occur fast enough to explain these patterns. The measured national differences are consistent with the environmental inputs that differ between rich and poor countries.
Partially — they correlate but are distinct.
PISA measures academic achievement in mathematics, reading, and science. IQ tests — particularly matrix reasoning variants — measure abstract fluid reasoning with minimal curriculum dependency. The two correlate moderately at the country level (high-IQ countries tend to have high PISA scores), but they diverge in instructive ways: Estonia scores higher on PISA than its IQ estimate implies, while some countries with moderate IQ estimates score below what wealth would predict. PISA is a useful independent benchmark precisely because it is a different measurement with different biases.
The data and claims in this article draw on the following primary and secondary sources.
Becker-Lynn National IQ Dataset
viewoniq.org — the primary academic database for country IQ estimates, updated 2019
OECD PISA 2022 Results
Official PISA 2022 Volume I — full international math, reading, and science rankings
Our World in Data: Global Education
ourworldindata.org — visualizations of global learning outcomes and schooling access
World Bank Human Capital Index 2024
worldbank.org — tracks education, health, and nutrition as joint determinants of productivity
According to the Becker-Lynn 2019 dataset, Japan and Taiwan share the top position with estimated average IQs of approximately 106.5, followed closely by Singapore (105.9) and Hong Kong (105.3). The exact ranking varies by dataset and methodology — Worlddata.info, which incorporates PISA data, places some countries differently. No ranking should be treated as a definitive final word.
Reliability varies enormously by country. High-income countries with multiple modern, nationally representative cognitive studies produce more reliable estimates. Many developing-world estimates rest on a single study, a regional sample, or data from the 1970s-1990s. Researcher Jelte Wicherts found that including all available African data raised the estimated sub-Saharan Africa average from about 70 to about 82 — a difference large enough to change policy conclusions entirely. Always check sample quality before trusting any specific figure.
Measured differences largely reflect environmental and methodological factors: years of schooling (each year adds roughly 1-3 IQ points), iodine and nutrition status (severe deficiency is associated with a 13.5-point average gap), lead exposure (population-level losses of up to 7 points in heavily exposed cohorts), test format familiarity, language translation quality, and sampling biases. These are changeable conditions, not fixed national traits.
The Flynn Effect describes the well-documented rise in raw IQ scores across many countries during the 20th century — roughly 3 IQ points per decade. Most researchers link this to improvements in nutrition, schooling, reduced disease burden, and test familiarity. Since the 1990s, some high-income countries including Norway, Denmark, Finland, and the UK have reported flat or slightly declining trends. The pattern is not universal: many developing nations continue to show gains as educational access improves.
Only loosely. Your IQMog score is an individual matrix-reasoning result from one online session. Country averages combine many different tests, years, languages, and populations into a single estimate. Your score is most useful for understanding your own percentile and reasoning baseline — not for drawing country-level conclusions. The within-country standard deviation is approximately 15 IQ points, meaning individuals within any country span a range far wider than the differences between most national averages.
What Is the Average IQ Score?
How 100 is defined, what the global mean looks like, and how norms are set.
IQ Score Ranges Explained
Every band from below 70 to 130+, with percentile context and interpretation notes.
How Accurate Are Online IQ Tests?
Test-retest reliability, norm quality, and what online scores can and cannot tell you.
Full IQ Score Chart
A complete score-to-percentile reference chart for all major IQ bands.
Free IQ Test
Take a Raven-style matrix reasoning assessment. No email or account required to start.
Mensa IQ Score Threshold
What score qualifies for Mensa and what that percentile actually represents.
Country averages are aggregate research estimates. Your personal reasoning baseline requires your own data. IQMog delivers a structured Raven-style matrix reasoning assessment and a paid results report with score, percentile band, and cognitive range explanation after completion. No account required to start.