Age at menarche is one of the few established risk factors for breast cancer; identification of its exogenous determinants could throw light on the origins of breast cancer. We have undertaken an epidemiologic study in Greece to ascertain whether: 1) energy intake, an indicator of physical activity, is associated with later age at menarche; 2) energy‐adjusted fat intake is related to earlier age at menarche; and 3) other macronutrients and anthropometric variables are predictors of age at menarche. Anthropometric, socio‐economic, familiar, nutritional and lifestyle predictors of age at menarche were studied by interviewing in person 345 female students 9 to 16 years old attending 8 schools of Greater Athens. Menarche was the outcome variable in a proportional hazards model assessing the mutually adjusted incidence rate ratio by a series of predictor variables. In a complementary analysis, age at menarche was the dependent variable among menstruating girls. Consistent results were derived from the main and the complementary analysis. Increased height and body mass index accelerate the occurrence of menarche. Maternal and daughter's ages at menarche are correlated, but there is no evidence of an association with paternal education. Various measures of moderate physical activity as well as increased total energy intake were associated with a delay in age at menarche. Energy‐adjusted macronutrients were not associated with age at menarche. It appears that an alteration of energy balance in early life through increased physical activity could delay age at menarche and reduce the risk for breast cancer in later life. © 1996 Wiley‐Liss, Inc.