BMI is a rough population-level tool that works poorly at the individual level. It correctly identifies obesity only about 43% of the time when checked against actual body fat measurements, while being very good at confirming that people who aren’t obese really aren’t (96% accuracy on that side). In other words, if BMI says you’re obese, it’s almost certainly right. But it misses more than half of people who actually carry excess body fat.
That gap between what BMI catches and what it misses explains why the American Medical Association adopted a policy stating that BMI “loses predictability when applied on the individual level” and should be used alongside other measures like waist circumference, body composition, and metabolic markers.
What BMI Actually Measures (and Doesn’t)
BMI divides your weight in kilograms by your height in meters squared. That’s it. It doesn’t distinguish between muscle, fat, bone, or water. A 200-pound person who is 5’10” gets the same BMI whether that weight comes from body fat or lean tissue. This makes BMI a useful shorthand for large-scale research, where individual variation averages out, but a blunt instrument in a doctor’s office.
When researchers compare BMI against body fat measured by DXA (a precise body-composition scan), the overall correlation is strong in groups. In children and adolescents, BMI correlates with fat mass at about r = 0.79 to 0.94 depending on how fat is expressed. But correlation across a population doesn’t mean accuracy for any one person. A BMI of 30 has a sensitivity of just 36% in men and 49% in women for detecting true obesity. That means BMI gives a passing grade to the majority of men and roughly half of women who actually have excess body fat by objective measurement.
Where BMI Gets It Wrong
The most familiar example is muscular people. In one study of college athletes, 38 had a BMI of 25 or higher, flagging them as overweight or obese. Only four of those 38 actually had excess body fat. Twenty-seven had high muscle mass instead. BMI simply cannot tell the difference between a linebacker and a sedentary person of the same height and weight.
The reverse problem is less well known but arguably more dangerous. Some people carry a normal BMI while harboring metabolically risky levels of body fat, particularly visceral fat packed around internal organs. BMI correlates better with the fat stored just under your skin than with visceral fat. Waist circumference, when added to BMI, significantly improves the ability to estimate visceral fat. Within every BMI category, people with larger waist circumferences carry more visceral fat, which is the type most strongly linked to heart disease and diabetes.
BMI Accuracy Varies by Ethnicity
Standard BMI thresholds were developed primarily from data on white European populations. For people of Asian descent, those cutoffs significantly underestimate risk. Studies show that young Asian Indians with a BMI around 23 (technically “normal”) already carry more total abdominal and visceral fat and process blood sugar less efficiently than white individuals at the same BMI. One hypothesis is that South Asian populations exhaust their capacity to store fat safely under the skin at lower overall body weights, leading fat to accumulate in and around organs sooner.
This isn’t a minor academic distinction. The World Health Organization proposed lowering the overweight threshold to a BMI of 23 for Asia-Pacific populations, and the American Diabetes Association now recommends screening Asian Americans for diabetes starting at a BMI of 23 rather than the usual 25. If you’re of South Asian, East Asian, or Southeast Asian background, a “normal” BMI may be providing false reassurance.
BMI Becomes Less Reliable With Age
As people age, they typically lose muscle mass while gaining fat, a shift that BMI is blind to. Someone whose weight stays stable from age 40 to 70 may have a very different body composition at 70, with significantly less muscle and more fat, yet their BMI remains unchanged. This matters because the combination of low muscle mass and high fat (sarcopenic obesity) carries serious health risks. People with sarcopenic obesity and two altered components of body composition face nearly three times the 10-year mortality risk, independent of their BMI.
In older adults, BMI also fails as a mortality predictor in a way it doesn’t in younger populations. Research on high-functioning older adults found no association between BMI and death risk in either unadjusted or adjusted analyses. Waist-to-hip ratio, by contrast, did predict mortality, with a graded relationship in women where each 0.1 increase raised death risk by 28%.
BMI Accuracy in Children
BMI-for-age percentiles are the standard screening tool for children, but their accuracy depends heavily on where a child falls on the chart. At the extremes, BMI works reasonably well: 93% of children at or above the 97th percentile truly had elevated body fat, and about 77% of those at or above the 95th percentile did too.
The “overweight” zone between the 85th and 94th percentiles is where BMI gets unreliable. Among 200 children in that range, roughly half had a moderate level of body fat, 30% had normal body fat, and only 20% had elevated body fat. A child in this zone could be carrying extra fat or extra muscle, and BMI alone can’t distinguish between the two. Misclassification rates were highest among Black children, with 50% of those in the 85th-to-94th percentile range actually having normal body fat levels. Researchers have described BMI as “almost useless as an estimator of percentage of body fat in normal-weight children.”
Metabolically Healthy Obesity
About 15% of adults classified as obese by BMI show no metabolic abnormalities: normal blood pressure, normal blood sugar, normal cholesterol, and normal triglycerides. This group, described as having metabolically healthy obesity, has grown from roughly 11% to 15% of the obese population between 1999 and 2018. Whether these individuals face long-term health risks comparable to metabolically unhealthy obesity remains debated, but their existence highlights that a BMI over 30 doesn’t automatically mean metabolic disease is present.
Better Ways to Assess Body Composition
No single number captures health risk perfectly, but several measures outperform BMI in specific ways. Waist circumference captures abdominal fat more directly. Among older adults, waist-to-hip ratio predicts mortality where BMI does not. Waist circumference also adds meaningful information on top of BMI: within every BMI category, a larger waist predicts more visceral fat.
The AMA now recommends combining BMI with measures like waist circumference, body composition assessment, and metabolic markers such as blood sugar and cholesterol. This layered approach catches what BMI misses: the thin person with dangerous visceral fat, the muscular person mislabeled as overweight, and the older adult whose stable weight masks a risky shift from muscle to fat.
For practical purposes, BMI still works as a rough screening tool for populations and as a starting point in clinical settings. It’s inexpensive, easy to calculate, and correlates with body fat in most people. But treating it as a diagnosis rather than a first pass is where it fails. If your BMI puts you in an unexpected category, whether that’s overweight when you’re visibly lean and active, or normal when you carry weight around your midsection, the number is worth questioning with more precise tools.