We have examined the association between induced or spontaneous abortion and breast cancer risk in Greece. In a hospital-based case-control study in Athens, 820 patients with confirmed breast cancer were compared with 795 orthopedic patient controls and 753 healthy visitor controls, matched to cases by age and interviewer. Logistic regression was used to analyze the data, controlling for demographic, reproductive and nutritional variables. Odds ratio (OR) patterns were similar for the 2 control series, which were therefore combined to increase precision of the estimates. The risk for breast cancer was not increased for women who had a history of abortion, compared to nulliparous women with no histoty of abortion. Thus ORs and 95% confidence intervals were for nulliparous women with spontaneous abortion, I . I7 (0.64-2.13): for nulliparous women with induced abortion, 0.98 ( 0 . 5 6 I .73); for parous women with no abortion, 0.56 (0.31-1.01); for parous women with spontaneous abortion, 0.6 I (0.33-I . 14) and for parous women with induced abortion, 0.99 (0.561.74). When the analysis was restricted to parous women, using parous women with no history of abortion as the baseline, ORs and 95% confidence intervals were for induced abortion before first full-term pregnancy, 2.06 ( I .45-2.90); for induced abortion after first full-term pregnancy, I .59 (I .24-2.04) and for spontaneous abortion, I . I0 (0.82-1.40). Our findings suggest that an interrupted pregnancy does not impart the long-term protective effect of a full-term pregnancy attributable to terminal differentiation.