The occurrence and course of eating disorder in a large representative population of 15-year-old London schoolgirls has been assessed using a two-stage survey methodology. Attempts to control weight were common and often transient. Dieting was in the great majority of girls found to be a benign practice without progression to more extreme concerns about food and weight. However, a small proportion of dieters did become cases and formed the majority of new cases found at follow-up. The relative risk of dieters becoming cases was eight times that of non-dieters. Many factors conventionally associated with eating disorder were associated more with attempting weight control than caseness. These included pre-morbid personality, pre-morbid obesity and family weight pathology. Other factors, including social class, career choice and psychosexual development, had no association either with attempting weight control or caseness.