Age at menarche, an early determinant of breast cancer risk, shows both a downward secular trend and marked differences by socioeconomic status, both presumably the reflection of dietary variations. A cohort study is being conducted in the Quebec City, Canada, area to assess the relation between diet, physical activities, and menarche. In the fall of 1986, 3,022 fifth-grade girls (mean age, 10.7 years) agreed to participate in the study. All of the girls' parents provided information on health, familial, and socioeconomic variables. A 3-day dietary record and a 7-day physical activity recall were provided by 95.8% of the girls. All dietary information was verified with each girl by a dietician. Body weight and height and two skinfold thickness measurements were also recorded for 98% of the subjects. At the beginning of the study, 107 girls had already experienced menarche. In the fall of 1987, letters were sent to the mothers of the remaining 2,915 girls; 2,854 (97.9%) responded. In the preceding year, 352 girls had reached menarche. Of these, 333 (94.6%) had provided baseline dietary data. A case-control analysis was conducted in which these 333 girls were compared with 333 age-matched premenarcheal girls. There was a weak association between energy intake, energy expenditure, and early menarche. Weight, height, skinfold thickness, mother's age at menarche, and participation in a dance, ballet, gymnastics, or figure skating club were also associated with the early onset of menstruation.