OBJECTIVE Treatment recommendations for localized prostate cancer may be improved by the identification of tumor factors prognostic for local control and survival. In this retrospective study, flow cytometric deoxyribonucleic acid (DNA) ploidy analysis and cell cycle analysis were performed on paraffin-embedded biopsy material to determine if additional prognostic factors could be identified in patients treated with radiation therapy. METHODS Seventy patients with T1-4NxM0 tumors were identified in whom the primary treatment had been radical radiation therapy with no prior or concurrent endocrine therapy and in whom sufficient prostatic tissue was available for flow cytometric analysis. There were 40 diploid, 26 aneuploid, and 4 multiploid cases. Aneuploid and multiploid cases were combined for analysis. Cell cycle data were obtained on all diploid and 10 aneuploid cases. RESULTS The histologic differentiation of the tumor (well or moderate vs. poor) was an independent predictor of overall survival and disease-free survival (p = 0.05 and 0.01, respectively). Local control was worse in the poorly differentiated patients, although this was not statistically significant in a multivariate analysis (p = 0.08). Neither T-stage, deoxyribonucleic acid ploidy (diploid vs. nondiploid), percent S-phase fraction, nor total proliferative fraction (S-phase fraction + G2M) significantly predicted for any of these endpoints. Within the diploid and well or moderately differentiated subgroup (n = 25), S-phase (< 4.2 vs. > or = 4.2) was a significant predictor of local control (100% vs. 51%, p = 0.03). A comparable distinction could be made using total proliferative fraction (< 10% vs. > or = 10%) with local control rates of 100% vs. 56% (p = 0.05). Among the poorly differentiated tumors, no similarly favorable subgroup was identified. CONCLUSIONS This retrospective and multivariate analysis identifies both histology and percent S-phase or total proliferative fraction as predictors of local control following irradiation, and confirms that histology, but not DNA ploidy, is significant for overall survival. If these previously unreported findings are confirmed by prospective studies, S-phase should be added to histology as a parameter in the evaluation of clinical trials.