A multiethnic cohort of 1378 Southern California school girls aged 8-13 years was followed for 4 years to evaluate factors predicting age at menarche, a risk factor for breast cancer. Height and weight were measured and dietary intake was assessed using a semi-quantitative food frequency questionnaire. Of 939 girls providing data on menarcheal status, 767 were premenarcheal at the start of the study; 679 girls provided acceptable dietary data and were included in the analyses. Cox proportional hazards models were used to assess the relationship between diet, body size, ethnicity and age at menarche. Hispanic, Asian/Pacific Island and African-American girls were more likely to experience early menarche than non-Hispanic white girls. Tall (> 148.6 cm) versus short (< 135.9 cm) girls experienced earlier menarche (relative hazard (RH) = 2.9, 95% confidence interval (CI) 2.1-4.1) as did those with high Quetelet's index (QI, kg m(-2)) (> 20.7) versus low QI (< 16.1) (RH = 2.2, 95% CI 1.7-2.9). Of all the dietary variables analysed, only energy intake was related to age at menarche. High versus low energy intake (> 12,013 kJ vs < 7004 kJ) was associated with a delay in menarche (RH = 0.7, 95% CI 0.5-0.9); this finding was limited to a subset of heavy Hispanic girls who appeared to underreport their dietary intake.