Cellular DNA content and proliferation rates have been suggested as prognostic factors in breast carcinomas. A series of 271 lymph-node negative breast carcinoma patients without adjuvant therapy was reviewed (mean follow-up, 108 mo). Tumor cells from the same paraffin-embedded block tissue (Hedley's method) were analyzed by image analysis (IA) in Feulgen-stained smears and by flow cytometric analysis (FC). Clinicopathologic features, ploidy, S-phase fraction, and percentage of tumor cells with more than 5n DNA content (in diploid tumors, by IA) were related to outcome. The results of IA and FC were compared in 115 cases. Tumor size, histologic grade, desmoplasia and S-phase fraction were significant predictors of survival in multivariate analysis (Cox proportionate regression) (P < or = 0.03). Ploidy by the two methods showed agreement in 100 carcinomas (87%). Of the 15 discordant cases, FC detected 6 multiploid and 4 aneuploid-peridiploid. In contrast, IA detected more tetraploid carcinomas. Tumor size, histologic grade, desmoplasia, and S-phase fraction were independent predictors of long-term prognosis in our patients. Ploidy and percentage of tumor cells with more than 5n DNA content were not prognostic indicators. FC detected aneuploidy more frequently than did IA.