This case-control study was designed to reevaluate the association of the morphology of breast tissue seen on mammograms with breast cancer risk and to assess the relation of diet, especially intake of fat and vitamin A, to the high-risk mammographic images. The cases included 290 patients with newly diagnosed breast cancer who were first treated in Quebec in 1982-1984. The controls included 645 women who participated in the Canadian National Breast Screening Study. Risk of breast cancer was higher among women with the P2 or DY parenchymal pattern (relative risk (RR) = 3.7, 95% confidence interval (Cl) 2.0-7.0) than it was among those with the N1 pattern. Moreover, risk increased regularly with the extent of nodular and homogeneous densities on the mammogram. Relative risk was 5.5 (95% Cl 2.3-13.2) for women in whom 60% or more of the volume of the breast showed either nodular or homogeneous densities compared with women without such densities. Among controls, increase in energy-adjusted saturated fat intake was associated with an increase in extent of high-risk mammographic features. Energy-adjusted polyunsaturated fat or cholesterol intake did not, however, appear to influence the morphology of breast tissue seen on the mammogram. Increasing carotenoid and fiber intakes were associated with a reduction of the extent of densities on the mammogram, but retinol intake seemed to have little or no effect on mammographic features. These data suggest that elevation in saturated fat intake and reduction in carotenoid and fiber intakes may be related to an increase in breast cancer risk through effects of these nutrients on breast tissue morphology.