In a hospital-based case-control study of common malignant epithelial tumors of the ovary, conducted in Athens (1989-1991), 189 cases were compared with 200 hospital visitor controls. Personal interviews were conducted in all cases and diet was ascertained through a semi-quantitative food frequency questionnaire. Nutrient intakes for individuals were estimated by multiplying the nutrient content of a typical portion size for each specified food item by the frequency at which the food was consumed per month and summing these estimates for all food items. Data were analyzed using logistic regression, controlling for non-dietary confounding factors, total energy intake and, among nutrients, mutual confounding influences. Adjusted odds ratios (rate ratios) for ovarian cancer, associated with particular nutritional variables, were expressed in terms of increments approximately equal to the standard deviations of (the residual of) the respective nutrients, on a daily basis. The adjusted odds ratios (and 95% confidence intervals) were 0.80 (0.65-0.99) for mono-unsaturated fat and 0.73 (0.61-0.87) for crude fiber. No substantial, statistically significant or consistent independent associations were noted for total energy, total protein, saturated fat, polyunsaturated fat, dietary cholesterol, total carbohydrates, sucrose, vitamin C, vitamin A, riboflavin or calcium. These associations, if causal, could explain to some extent the relatively low incidence of ovarian cancer in Greece and other Mediterranean countries as well as the increasing incidence trends noted in these countries during the last few decades.