To be more effective, the prevention of obesity in childhood should be focused on the population at risk. The purpose of the present study is firstly to find correlations between certain environmental factors and obesity in childhood, and secondly to measure the influence of the environmental factors after taking the parental history of obesity into account. This case controlled study includes 704 controls vs. 327 cases selected in a population of five year old school children. The anthropometric assessment was completed at school. Obesity was defined as a weight for height > or = 2 s.d. using the French weight charts for French children based on sex and height. Interviews of the parents recorded parental overweight and child birth overweight as 'constitutional' factors and family structure, socio-economic level and daily lifestyle (sleep, TV viewing, after school care, etc.) as 'environmental' factors. The results show that parental overweight and birth overweight are closely related to the child's obesity at five years of age (estimated relative risks 3.1 and 2.4 respectively). The environmental factors which contribute to child obesity are: southern European origin of the mother, snacks, excessive television viewing and, more importantly, short sleep duration (estimated relative risks = 1.9, 1.3, 2.1 and 4.9 respectively). A logistic regression model, after taking parental overweight into account, shows that the relationship between obesity and short sleep duration persists independently of television viewing. The hypotheses raised by these findings are discussed.