This study examines the cross-sectional relationships between diet and serum cholesterol in a sample of 413 Framingham cohort males, ages 37-70. Regression analyses controlling for age, calories, and coronary heart disease risk factors showed that the direct predictors of serum cholesterol included: fat (g) (p less than 0.05), cholesterol (mg/1000 calories) (p less than 0.05), protein (g and calorie density) (p less than 0.05 and p less than 0.001, respectively), Metropolitan relative weight (p less than 0.05), and systolic blood pressure (p less than 0.001). Fat (calorie density) and cholesterol (mg) were marginally significant direct positive predictors (0.05 less than or equal to p less than or equal to 0.10). Inverse correlates of serum cholesterol were total carbohydrate (g and calorie density) and simple carbohydrate (calorie density) (p less than 0.001). A marginally inverse correlate of serum cholesterol was complex carbohydrate (calorie density) (0.05 less than or equal to p less than or equal to 0.10). These cross-sectional results show that dietary variables predict serum cholesterol levels in men independent of other coronary heart disease risk factors.