Data collected between 1973 and 1984 on 696 incident cases of breast cancer and 1,376 matched controls from four Breast Cancer Detection Demonstration Project clinics in the United States were used to assess the role of mammographic parenchymal pattern as a risk factor and its relationship with other, accepted, risk factors. The data confirm previous reports of the influence of benign breast biopsy, age at first live birth, family history of breast cancer, and duration of menstruation on the incidence of breast cancer. Height is also found to be an influential factor. Parenchymal pattern is found to be a risk factor with effects comparable in magnitude to the other factors studied. It operates separately from them, except for its relationship with height and weight. After adjustment for parenchymal pattern, weight is seen to have a significant effect on breast cancer incidence, and height is no longer needed in a model for risk. A model which simultaneously incorporates all of the risk factors considered, including parenchymal pattern, is presented. While these factors are of interest in the epidemiology of breast cancer, it is demonstrated that they are insufficient to allow reliable prediction of the disease in an individual woman.