Prostate-specific antigen (PSA) is a valuable tumor marker used for diagnosis and management of prostate cancer. Recently, PSA has been found in various female tissues and body fluids. Female breasts, both normal and abnormal, including cancerous tissues, can produce PSA, and this production is regulated by androgens and progestins. Preliminary data suggested that patients with breast tumors positive for PSA may have better prognosis compared to those with PSA-negative breast tumors. This study examines the prognostic value of PSA in a large cohort study of United States patients. Using a PSA assay that has a lower detection limit of 0.001 ng/ml, we measured PSA in tumor cytosolic extracts of 953 women with primary breast cancer. Other information available for this study included age, follow-up time, survival outcome, tumor size, nodal status, steroid hormone receptor levels, DNA analysis by flow cytometry, and postoperative treatment. The median follow-up time was 73 months. During the follow-up, 200 patients relapsed and 188 died. PSA presence was found to be significantly associated with smaller tumors, tumors with low S-phase fraction, diploid tumors, younger patient age, and tumors with lower cellularity. Survival analysis indicated that the relative risks (RRs) for relapse and death were both significantly lower [RR = 0.67 (P = 0.01) for relapse; RR = 0.72 (P = 0.05) for death] in PSA-positive patients (levels higher than the 30th percentile of PSA values) than in PSA-negative patients. The reduced risks for relapse and death remained statistically significant after other clinical and pathological variables were adjusted in the multivariate analysis [RR = 0.68 (P = 0.02) for relapse; RR = 0.65 (P = 0.02) for death]. Our results suggest that the measurement of PSA in breast tumor extracts provides additional information on the prognosis of patients with primary breast cancer.