Obesity and pattern of fat distribution are both important factors related to poor health outcomes. Many measures of obesity and fat distribution pattern have been employed by different authors and to facilitate interpopulation comparisons and interpretation of secular trends it is necessary that standardized methods for measurement and classification are set in place. The use of BMI as a measure of fatness for epidemiological studies is widely accepted, easily measured and BMI predicts morbidity and mortality in many populations. The most appropriate level at which to define obesity is a matter of debate but systems which use BMI ≥ 25 and ≤ 30 kg/m2 as overweight, and BMI > 30 kg/m2 as obese for all adults are simple, easily remembered, already widely used and BMIs above 30 kg/m2 are clearly associated with increased risk of morbidity and mortality. In some populations there may be a case for using a lower cut-off but not unless there is specific evidence to support this. For the present WHR is probably the best method for assessing fat distribution, although waist circumference on its own may be more useful in determining risk levels. Standard sites for measurement of both waist and hip girths have been described. There is a large variation in the prevalence of obesity across the populations for which data is available, with high prevalences of obesity and dramatic secular trends especially apparent in modernizing Pacific Island populations. The 'thrifty genotype' hypothesis has been invoked to try and explain this situation. The clustering of obesity, NIDDM and CVD risk factors has been recognized and various 'syndromes' have been described which group different factors together, with hyperinsulinaemia and insulin resistance proposed as the underlying problem.