The Greek diet is characterized by a high total fat but low saturated fat intake. In a hospital-based case-control study of female breast cancer conducted in Athens (1989-91), 820 patients with confirmed cancer of the breast were compared with 795 orthopaedic patient controls and 753 hospital visitor controls, matched to the cases by age and interviewer. Diet was ascertained through a semiquantitative food frequency questionnaire; macronutrient intakes were estimated from the nutrient content of a selected typical portion size for each specified food item, summed for all items. Logistic regression was used to analyse the data, controlling for demographic and reproductive risk factors for breast cancer as well as for total energy intake and mutual confounding influences among nutrients. There was no significant or suggestive association of total protein, total fat, categories of fat or total carbohydrates with breast cancer risk. Thus, the mutually adjusted relative risk per quintile and (in parenthesis) 95% confidence interval were: for protein, 1.06 (0.94-1.20); saturated fat, 0.99 (0.89-1.11); monounsaturated fat, 0.97 (0.88-1.07), polyunsaturated fat, 1.05 (0.97-1.13); and total carbohydrates, 1.03 (0.94-1.12). In alternative analytical approaches only total protein appeared to be positively associated to the occurrence of breast cancer with some consistency, but the results were far from statistically significant. These findings do not support a role for fat or other energy-generating nutrients in the aetiology of breast cancer.