To evaluate the relationship between breast cancer risk and spontaneous and induced abortion, we conducted a detailed descriptive review of 32 epidemiologic studies that provided data by type of abortion and by various measures of exposure to abortion-number of abortions, timing of abortion in relation to first full-term pregnancy, length of gestation, and age at first abortion. Breast cancer risk did not appear to be associated with an increasing number of spontaneous or induced abortions. Our review also suggested that breast cancer risk probably was not related to the other measures of exposure to abortion, and probably did not differ by age or a family history of breast cancer. Finally, the data appeared to suggest a slightly increased risk among nulliparous women, but this tendency was based primarily on studies with a small number of nulliparous women who had had spontaneous or induced abortions. Definitive conclusions about an association between breast cancer risk and spontaneous or induced abortion are not possible at present because of inconsistent findings across studies. Future investigations should consider prospective designs, separate analyses of spontaneous and induced abortions, appropriate referent groups, and adequate adjustment for confounding and effect modification. Future investigations also should attempt to determine whether any increased risks reflect the transient increase in breast cancer risk hypothesized for full-term pregnancy or a causal relationship specific to spontaneous or induced abortion.