We are conducting a long-term randomized controlled trial to determine if intervention with a low-fat high-carbohydrate diet reduces breast cancer risk. The present study examines the effects of 2 years of dietary intervention on serum sex hormone levels in premenopausal women. Subjects with extensive mammographic densities were enrolled in a dietary intervention trial. The intervention involved intensive individual counselling aimed at reducing total dietary fat to 15% of calories. Control subjects received general advice about diet but were not counselled to change their fat intake. Serum sex hormone levels were measured in 220 premenopausal subjects at entry and 2 years after randomization. Two years after randomization oestradiol levels were 20% (70 pmol l(-1)) lower (P = 0.04) and progesterone levels were 35% (1.0 nmol l(-1)) lower (P = 0.004) and follicle-stimulating hormone (FSH) levels were 7% (1 IU) higher (P = 0.38) in the intervention group than in the control group. The FSH-oestradiol ratio was 13% higher in the intervention group (P = 0.18). Samples analysed accounting for the timing of the blood sample in relation to the menstrual cycle showed that, in the intervention group, oestradiol and progesterone levels were lower and FSH levels higher in subjects with blood samples taken more than 30 days after the last menstrual period. Because of the strong evidence linking ovarian hormonal activity to breast cancer risk, these results suggest that a low-fat high-carbohydrate diet may reduce risk of breast cancer by reducing exposure to ovarian hormones that are a stimulus to cell division in the breast.