Different mammographic parenchymal patterns have been found to be associated with significantly different rates for the development of breast cancer in a screening program of self-referred women. These differences are qualitatively similar but of lesser magnitude than those in previous reports by Wolfe which were based on symptomatic women who had had a previous negative mammogram. In addition, this report indicates a small difference in the rate of breast cancer at first mammographic examination, using the same parenchymal classifications. These findings, coupled with other risk factors, may permit the concentration of mammographic screening on a smaller segment of the population at risk, thus improving the benefit to risk ratio.