The epidemiological evidence that a high-fat diet promotes the development of obesity is considered suggestive but not definitive. The purpose of this paper is to provide a review of various epidemiological methods that have been used to address this issue as well as an updated summary of the existing evidence. Ecological studies describing dietary fat intake and obesity at the population level provide mixed results and are likely to be biased by both confounding and unknown data quality factors that differ systematically across the populations studied. Cross-sectional studies are generally in agreement that the concentration of fat in the diet is positively associated with relative weight. Prospective studies of diet in relation to subsequent weight change give inconsistent results. This may be due to behavioural factors such as dieting in response to weight gain; in addition, this type of study rarely takes into account the possible interaction between genetic predisposition and dietary fat in promoting weight gain. Finally, intervention studies in free-living subjects are considered, providing evidence of a consistent but short-lived period of active weight loss on low-fat diets. The experimental evidence on this relationship is more conclusive than the epidemiological evidence, although biological mechanisms remain controversial. Some areas for future epidemiological research involve: longitudinal studies of dietary fat intake as a predictor of growth in children; observational studies relating total dietary fat and specific types of fat to overall as well as regional adiposity; and randomized intervention studies of the effect of low-fat diets with particular emphasis on and familial predisposition to obesity and other possible modifying factors.