Most geneticists and biological anthropologists agree that discrete biological human races do not exist. Humans share roughly 99.9% of their DNA, and the genetic differences that do exist between populations are far smaller than those found within any single population. That does not mean all humans are genetically identical or that geography has left no mark on our genomes. It has, quite visibly. But the pattern of that variation looks nothing like the neat racial boxes people have used for centuries, and understanding why matters for medicine, justice, and how you interpret your own ancestry results.
Where the Idea Came From
The modern concept of biological race traces largely to the work of Johann Friedrich Blumenbach, an eighteenth-century German anatomist. Blumenbach divided humanity into five “varieties” based on skull shape and other physical features, placing Europeans at the top. His categories were not the product of careful measurement or statistical analysis. He chose a skull from the Caucasus mountains as the ideal representative of Europeans because he found it beautiful, and named the group “Caucasian” on that aesthetic basis alone. The entire framework rested on subjective judgments about which skulls and which bodies looked most appealing to an eighteenth-century European professor.1Europe PMC / BMJ. The beautiful skull and Blumenbach’s errors: the birth of the scientific concept of race
This matters because Blumenbach’s categories became the scaffold for two centuries of racial science. Subsequent researchers layered anthropometric measurements, intelligence tests, and medical assumptions on top of a system that was never grounded in objective biology to begin with. The categories persisted not because they kept proving themselves in data, but because they had become embedded in legal systems, social hierarchies, and institutional practices that made them feel natural.
What the Genetics Actually Shows
The most influential genetic challenge to biological race came in 1972, when the evolutionary biologist Richard Lewontin analyzed protein variation across human populations. He found that about 85% of all human genetic diversity exists within any single population. Only around 6% to 7% of total genetic variation falls between groups traditionally called races, with another 8% or so separating populations within those groups. In Lewontin’s own words, the result was “quite remarkable” for how little genetic difference it showed between so-called races.2Philosophical Transactions of the Royal Society B: Biological Sciences. Celebrating 50 years since Lewontin’s apportionment of human diversity
Lewontin’s finding has been replicated and refined with much larger datasets over the past half century. The numbers shift slightly depending on the genetic markers used and how populations are defined, but the core message holds: if you picked two people at random from different continents, the genetic difference between them would be only modestly greater than the difference between two people from the same village. That is not what you would expect if humanity were divided into distinct biological types.
Critics have pointed out that even small between-group differences, when spread across thousands of genetic loci, can produce patterns that a clustering algorithm can detect. Software like STRUCTURE can group individuals into clusters that correspond loosely to continental ancestry when asked to do so.3Europe PMC. An overview of STRUCTURE: applications, parameter settings, and supporting software But the number of clusters these programs find depends entirely on how many clusters the researcher tells the software to look for. Ask for two clusters and you get a split along one axis. Ask for seven and you get a different map. The clusters are real statistical patterns, but they are not evidence that humanity naturally divides into a fixed number of biological races. They are a reflection of how geography, migration, and isolation have shaped allele frequencies over thousands of years.
Geography Does the Heavy Lifting
The pattern that genetics reveals is not one of discrete groups separated by sharp boundaries. It is one of gradients. Genetic differentiation between human populations increases smoothly with geographic distance, and within-population diversity decreases with distance from Africa. These gradients, called clines, can explain most of the genetic variation observed across the globe.4Trends in Genetics. Going the distance: human population genetics in a clinal world
Think of it like an accent map. Someone from London sounds different from someone from Edinburgh, who sounds different from someone from Dublin. You can hear the differences, and they track geography, but there is no sharp line where one accent ends and another begins. Travel the route and the speech changes gradually, town by town. Human genetic variation works the same way. Populations that are geographically closer tend to be genetically more similar, with no clear cutoffs that would let you draw a line between one “race” and the next.
The reason Africa shows the most within-population diversity is straightforward: anatomically modern humans originated there and have lived there the longest. Every migration out of Africa was carried out by a subset of people carrying a subset of genetic variation. The further a population’s ancestors traveled from Africa, the more genetic variation they left behind along the way. This is why African populations are the most genetically diverse on Earth, a pattern completely incompatible with grouping all Africans into a single biological “race.”
Skin Color, the Most Visible Red Herring
Skin pigmentation is probably the trait people associate most strongly with race, and it is also one of the clearest illustrations of why racial categories fail as biology. Skin color is an adaptation to ultraviolet radiation. Near the equator, where UV exposure is intense, natural selection favored dark, melanin-rich skin that protects against folate destruction and DNA damage. At higher latitudes, where UV is weaker, lighter skin evolved because it allows more efficient production of vitamin D from sunlight.5PubMed Central. Human skin pigmentation as an adaptation to UV radiation
The critical detail is that depigmented skin evolved multiple times independently through different genetic pathways. European and East Asian populations both have relatively light skin, but they got there through partly different sets of genetic changes. Meanwhile, deeply pigmented skin is found in populations across Africa, South Asia, Melanesia, and Australia that are not closely related to one another. The same skin color can arise from quite different genetic backgrounds, and closely related populations can have very different skin tones depending on where they live.6PubMed Central. The evolution of human skin pigmentation involved the interactions of genetic, environmental, and cultural variables
This is the problem with using any single visible trait to define race: the traits that look most racial are often the ones most subject to local environmental selection, meaning they track climate and latitude rather than deep shared ancestry. Two populations with similar skin colors can be genetically more different from each other than either is from a population with a very different skin tone.
What the Scientific Community Has Concluded
Surveys of professional anthropologists show a strong consensus: humans cannot be meaningfully subdivided into biological races. A large survey found that a supermajority of physical anthropologists rejected the idea that humans can be divided into discrete biological races, rejected the notion that races have fixed biological boundaries, and agreed that genetic ancestry is a better proxy for genetic relationships than racial categories.7PubMed Central. Anthropologists’ views on race, ancestry, and genetics
The consensus is not that human biological variation does not exist. It plainly does: people differ in lactose tolerance, susceptibility to certain diseases, drug metabolism, and thousands of other traits. The consensus is that this variation does not sort neatly into a handful of racial bins. Trait distributions overlap extensively. A trait common in one population almost always exists at some frequency in many others. No single trait is exclusive to any one population, and the package of traits that people mentally associate with a given race does not travel as a bundle through the genome.
Similarly, a systematic review of genetic studies that used “African ancestry” as a category found that while researchers frequently used self-reported race to define study populations, none of the studies provided a genetic explanation for why race should function as a genetic category. The Human Genome Project confirmed that humans are about 99.9% identical at the DNA level, and the remaining variation does not organize itself along racial lines.8Europe PMC. Race and genetics versus ‘race’ in genetics: A systematic review of the use of African ancestry in genetic studies
When Race Enters the Clinic
If race is not a valid biological category, why does it show up in medical formulas and treatment guidelines? The answer is partly inertia, partly statistical convenience, and partly that race, as a social category, correlates with real-world exposures and experiences that affect health. Disentangling what is biological from what is social is one of the messiest problems in modern medicine.
A well-known example is the estimated glomerular filtration rate, the standard clinical measure of kidney function. For years, the formula included a race correction factor that adjusted the score upward for Black patients, effectively making their kidneys appear healthier on paper. The rationale was based on an assumption about average muscle mass by race. In practice, the correction delayed nephrology referrals, kidney transplant evaluations, and treatments for Black patients by overestimating their kidney function.9Europe PMC / Delaware Journal of Public Health. The Case Against Race-Based GFR Major medical organizations have since moved to race-free kidney function equations, recognizing that the race adjustment was reinforcing disparities rather than correcting for biology.
Pharmacogenomics complicates the picture. Some genetic variants that affect how your body processes drugs do differ in frequency across populations grouped by race or ethnicity. A large analysis found numerous pharmacogenomic variants with substantial allele frequency differences between racial and ethnic groups, and predicted that these differences could translate into hundreds of additional adverse drug reactions per thousand treated patients in some minority populations.10Europe PMC. Race, Ethnicity, and Pharmacogenomic Variation in the United States and the United Kingdom That finding does not mean race is biological. What it means is that populations with shared geographic ancestry sometimes share higher or lower frequencies of particular drug-metabolism variants, and that until individual genotyping becomes routine, population-level data can sometimes serve as a crude stand-in. The goal, most researchers agree, is to replace race-based prescribing with direct genetic testing as quickly as possible.
The Polygenic Risk Score Problem
Polygenic risk scores, which estimate your genetic predisposition to conditions like heart disease, diabetes, or breast cancer by summing the effects of many small genetic variants, are a frontier of precision medicine. They also expose a serious equity gap rooted in ancestry. Because most of the large-scale genetic studies used to build these scores have been conducted on people of European descent, the scores work best for that population and perform substantially worse for everyone else.
Research has shown that polygenic scores derived from European-ancestry data perform worst in people of African ancestry, where median predictive accuracy was only about 42% of what it was for matched European-ancestry individuals. Performance was also lower, though less dramatically so, in South Asian and East Asian samples.11Nature Communications. Analysis of polygenic risk score usage and performance in diverse human populations The takeaway is that the scores are not revealing anything fundamental about racial biology. They are reflecting the fact that the research pipeline overrepresents one population, and the genetic variants, linkage patterns, and environmental contexts that matter for disease prediction differ across ancestries in ways that cannot be captured by training models on a narrow slice of humanity.12PubMed Central. Clinical use of current polygenic risk scores may exacerbate health disparities
If these scores are rolled out clinically without fixing the ancestry bias, the people who stand to benefit least are those who already experience the worst health outcomes. That is the opposite of what precision medicine is supposed to do.
How Racism Gets Under the Skin
One of the most important distinctions in this entire discussion is between race as biology and race as lived experience. Race may not be a coherent biological category, but racism is a measurable force that shapes biology. The health disparities observed between racial groups in countries like the United States are not primarily genetic. They are largely driven by social determinants of health: access to food, housing, education, employment, exposure to pollution, encounters with the criminal justice system, and the chronic stress of discrimination.
Research on COVID-19 disparities found that the greatest health inequities occur at the intersection of low socioeconomic status and historically disadvantaged racial groups. But the picture is more complex than simple poverty. Racial health disparities persist even among higher-income minority groups and after controlling for socioeconomic factors. Highly educated Black women, for instance, experience close to double the rate of infant mortality compared to highly educated white women, a gap that income and education alone cannot explain.13Wiley Online Library. Social Determinants of Health Factors for Gene–Environment COVID‐19 Research: Challenges and Opportunities
A growing body of neurophysiological research documents how racism-related stress accumulates in the body over time. Experiences of both explicit discrimination and chronic systemic disadvantage are associated with measurable changes in stress hormones, inflammatory markers, heart rate variability, and even brain structure and function. These physiological effects may help explain why racial health disparities persist even after controlling for income and insurance status: the biological toll of living in a racialized society is itself a health exposure.14PubMed Central. The neurophysiological consequences of racism-related stressors in Black Americans
Forensic Ancestry Estimation and Its Limits
Forensic anthropology offers what might seem like a counterexample to the “race isn’t biological” claim: forensic scientists do estimate the ancestry of skeletal remains, and they often get it right. But a closer look at the practice reveals something more nuanced. Ancestry estimation relies on patterns of cranial measurements and non-metric skull traits that correlate with geographic origin. These correlations exist, but they overlap considerably between populations. No single skull feature is found in only one population, and accuracy suffers in individuals with mixed ancestry, which is increasingly common in a globalized world.15PubMed Central. Evaluation of ancestry from human skeletal remains: a concise review
The traits forensic scientists measure are phenotypic features shaped partly by heredity and partly by environment, including diet and climate. They offer a probabilistic guide to geographic origin, not a diagnostic test for race. Many forensic anthropologists have moved toward language that reflects this, preferring terms like “estimated ancestry” or “population affinity” over racial labels, and increasingly supplementing skeletal analysis with genomic data when possible.
DNA Tests and the Reification of Race
Consumer ancestry tests from companies like 23andMe and AncestryDNA have brought population genetics into millions of homes. These tests are genuinely informative about geographic ancestry, tracing the continental origins of segments of your genome with reasonable accuracy. But research suggests they may also be reinforcing the very racial categories that genetics undermines.
A nationally representative survey experiment found that exposure to racial admixture testing significantly increased beliefs that Black and white people are essentially and fundamentally different, a pattern researchers call the “reification hypothesis.” By presenting ancestry information through a racial lens, the tests may inadvertently make race feel more biologically real than the underlying science supports.16Europe PMC. Direct-to-Consumer Racial Admixture Tests and Beliefs About Essential Racial Differences
The irony is thick. The same genomic revolution that demonstrated the absence of discrete biological races may, through its consumer products, be strengthening public belief in them. When a DNA test tells you that you are “43% West African” and “51% European,” it is reporting real statistical signals about where your ancestors likely lived. But the act of packaging that information into continental percentages can make it feel like a racial recipe, as though your genome is assembled from biologically distinct ingredients rather than drawn from a continuous web of human variation that happens to be structured by geography. This is one of the more consequential misunderstandings in popular science right now, and the testing companies have been slow to address it clearly.