The surface epithelial glycoprotein MUC1 becomes overexpressed and hypoglycosylated in adenocarcinomas; similar changes occur during nonmalignant inflammatory events. Antibodies developed against tumor-like MUC1 in response to such events could be one way through which ovarian cancer risk factors operate. We evaluated the association between anti-MUC1 antibodies and risk of ovarian cancer in a prospective nested case-control study in the Nurses' Health Studies. We used an ELISA to measure plasma anti-MUC1 antibodies in 117 ovarian cancer cases collected at least 3 years before diagnosis and 339 matched controls. In controls, younger women (P-trend = 0.03), those with a tubal ligation (P = 0.03), and those with fewer ovulatory cycles (P-trend = 0.04) had higher antibody levels. In cases, women with late-stage disease (P = 0.04) and those whose specimen was >11 years remote from diagnosis (P = 0.01) had higher antibody levels. Overall, increasing anti-MUC1 antibody levels were associated with a nonsignificant trend for lower risk for ovarian cancer, but there was highly significant heterogeneity by age (P-heterogeneity = 0.005). In women <64 years, the antibody level in quartiles 2 to 4 versus quartile 1 were associated with reduced risk (relative risk = 0.53; 95% confidence interval, 0.31-0.93; P-trend = 0.03), whereas in women > or = 64 years, the corresponding relative risk was 2.11 (95% confidence interval, 0.73-6.04); P-trend = 0.05). Anti-MUC1 antibodies evaluated several years before diagnosis may be associated with lower risk of subsequent ovarian cancer in women <64 years old at assessment. Key elements of an "immune model" to explain ovarian cancer risk factors are confirmed and should be evaluated in larger prospective studies.