To examine the effects of dietary fat intake on breast cancer risk, we are conducting a randomized trial of dietary intervention in women with extensive areas of radiologically dense breast tissue on mammography, a risk factor for breast cancer. Early results show that after 2 years on a low-fat, high-carbohydrate diet there is a significant reduction in area of density, particularly in women going through menopause. In women who went through menopause during the 2-year follow-up, the mean decreases in area of density and percentage of density in the intervention group were 11.0 cm2 and 11.0%, respectively, whereas the control group decreased 4.5 cm2 and 5.2%. The purpose of this analysis was to determine whether changes in intake of specific macronutrients could account for the observed reduction in breast density in these women. Differences between 2-year and baseline values of macronutrients (averaged over 3 nonconsecutive days of food intake) were calculated. We examined the effect of dietary variables, adjusted for changes in total calorie intake and weight and for family history of breast cancer, on changes in area of density and percentage of density using linear regression. Reduction in total or saturated fat intake or cholesterol intake was significantly associated with decreased dense area (p < or = .004). The most significant dietary variable associated with reduction in percentage of density was reduction in dietary cholesterol intake (P = 0.001), although reducing saturated fat intake was of borderline significance (P = 0.05). The effect of the membership in the intervention and control groups on change in area of density or percentage of density was reduced by models that included changes in intake of any fat, or cholesterol, or carbohydrates. The observation of an effect of diet at menopause on breast density, a marker of increased risk of breast cancer, may be an indication that exposures at this time have an enhanced effect on subsequent risk.