This study examined relationships between diet and plasma total and LDL cholesterol levels in a population-based sample of 695 premenopausal and 727 postmenopausal women participating in the Framingham Offspring/Spouse Study. Regression analyses controlled for age, caloric intake, apolipoprotein E isoform type, estrogen use, and important CVD risk factors indicated that plasma total and LDL-cholesterol levels were directly associated with consumption of saturated fat and inversely associated with total calorie intake. In contrast, dietary cholesterol was not a predictor of plasma total or LDL cholesterol levels. Total cholesterol levels were also directly associated with total fat, oleic acid, and animal fat, and inversely associated with carbohydrate intake. Stepwise regressions with key nutrients indicated that saturated fat was consistently associated with total and LDL cholesterol levels in Framingham women. These analyses suggest that diet explains 2% of the variability in these lipid levels in a cross-sectional sample of women; the full model explains 22-27%.