BMI, or Body Mass Index, is a simple formula using a person's height and weight to calculate a number which is supposedly representative of their level of body fat.
The formula itself is:
Weight (kg) / Height (m)2
(Weight (lbs) * 703) / Height (in)2
And the numerical value fits into the following categories:
Less than 18.5 is seen as underweight
18.5-24.9 is seen as average weight and is usually a target range
25-29.9 is seen as overweight
30 and above is seen as obese
In the obese category, it is further divided into 30-34.9 (Class I Obesity), 35-39.9 (Class II Obesity) and 40+ (Morbid Obesity). The first two classes are seen as at-risk populations whereas health complications associated with obesity are assumed in a state of Morbid Obesity.
BMI is not the only measurement of obesity, although it is the most commonly used.
Many studies report obesity rates as a function of body fat percentage (BF%), which is more accurate but requires special tools to assess; even then, the tools may not be highly accurate.
The standard for obesity as defined by body fat percentage is greater than 25% body fat for men and greater than 35% for females; this is the World Health Organization's (WHO) reference standard made in 1995. A more recent (2009) proposal from the American Society of Bariatric Physicians suggest lowering the values to 25% and 30% for men and women, respectively, and these values are used in some studies.
BMI tends to be used in large-scale population research and surveys as it can be calculated from height and weight, either self-reported or taken quickly and non-invasively by a researcher.
It has the benefits of being quick, easy to calculate, and a measure of body fat in which most of the population will consent to (unlike calipers which are invasive due to touching skin with cold metal objects, and hydrostatic weighing which dunks people in water without oxygen in their lungs).
An exception to this is the NHANES series of studies, a large scale series of studies conducted in the US that assesses persons by calipers.
In a study of 1,676 young girls (aged 5-16) it was found that although ethnicity differences existed in body fat that 89.9-92.4% of girls were accurately diagnosed with BMI. Results from NHANES 1999-2004 (three different NHANES surveys) found that 86.9%-89.1% of youth between the ages of 5-18 (both genders) were accurately diagnosed with BMI when compared against skin-fold calipers.
A cross-sectional study of 13,601 subjects in the US compared BMI against BIA (Bioelectrical Impedence Analysis). BMI defined 21% of men and 31% of women as obese, and BIA indicated 50% of men and 61% of women. Results from this study should be taken with a bit of scrutiny, as BIA is a measure of body fat with high variability based on hydration status.
A smaller scale study (1,691 persons) using DEXA scans (seen as a valid body fat measuring device) found that there was a 34.7% discrepancy between BMI and DEXA for women and 35.2% for men. However, BMI appeared to misclassify women as less fat as they were by DEXA; notable misclassifications include 20.3% of women being obese via BMI while DEXA showed 37.1%, 24.8% of men being obese via BMI compared with 38.4% of men being obese via DEXA. These results have been replicated in which persons in the normal BMI range were actually obese according to body fat percentage (20% of men, 9.2% of females) and more persons in the overweight BMI range were actually obese by body fat percentage (67.2% of men, 84.2% of females). High obesity rates in this study may be partially explained by socioeconomic issues, as it comprised Mexican persons (n=538) living in the southern USA. Finally, another study utilizing DEXA on a sample size of 1,393 persons found that 26% of persons were classified obese by BMI while 64% of persons were obese by DEXA; a misclassification rate (false negatives) of 25% for men and 48% for women was noted.
This trend of BMI underreporting obesity has been replicated in a small (82) sample of active police officers, and for child-bearing aged women where BMI suggests 36.9% of these persons are obese and the WHO standard of 35% body fat indicates an obesity rate of 63.1%.
One meta-analysis on the subject suggests that BMI fails to classify half of persons with excess body fat, reporting them as normal or overweight despite having a body fat percentage classifying them as obese.
If you are normal weight or overweight according to BMI (18.5-29.9) there is still a chance you are actually obese, and thus is primarily due to low levels of lean mass (muscle, water, and glycogen).
If you are obese according to BMI, you are most likely obese according to body fat percentage as well. When sampling from the general population, over 95% of men and 99% of women identified as obese by BMI were obese via body fat levels.
Outliers to this dataset, those who have enough lean mass to be classified as obese by BMI but not by body fat percentage, are far and few in society. These persons would normally be highly active athletes or dedicated 'weekend warriors', and it is unlikely sedentary persons or those with infrequent exercise habits would be these outliers.