BMI originated as a population measure. And it works extremely well as a population measure, in that there is a strong correlation between the incidence of >25 BMIs and the incidence of various diseases - CHD, diabetes, some cancers.

Inevitably, when a population measure is applied to individuals there will be anomalies. There are very muscular individuals whose BMI indicates that they are obese but who have low body fat. Many, many people believe they come into that category.

Unfortunately for most of them, such people are quite rare. The number of States in the USA with more than 20% of their population classified as obese has grown from zero to 50 in the last 30 years. Somehow, I find it difficult to believe that this has happened because the whole US population has been working out.

For most of us, BMI is indeed painfully accurate. My own is just about 25. I know from experience that my fighting weight is about 7kg (16lbs) lower, a BMI of 23. I'm a fairly heavily-built, broad-shouldered guy.