Body mass index (BMI) was first proposed in 1835 as a way to standardise body composition assessment for people of different heights, at a time when malnutrition was the main public health concern. BMI has been considered appropriately as a part of nutritional assessment in populations. It is not, however, a useful tool for assessment of individuals because there is so much individual variability in body composition and in its impact on health outcomes. Similarly, high BMI does not distinguish between excess body fat (bad for health) and large muscle mass (good). In contrast, we propose that individuals need to be assessed using clinical criteria, monitored over time to trigger different interventions. A diagnosis of obesity should be based on estimates of body fat (BMI, now being replaced by percentage body fat) at a particular age, and a clinical staging system.