BACKGROUND The biologic behavior of synovial sarcoma remains a matter of controversy. Some investigators considered proliferative activity to be an important prognostic factor in this tumor. METHODS Fifty-six patients with synovial sarcoma were immunohistochemically studied with PC10, a monoclonal antibody to proliferating cell nuclear antigen (PCNA). The percentage of nuclear areas with positive staining for PCNA (PCNA score), quantified by using an image analyzer, was compared with nuclear atypia, tumor necrosis, mitoses, and survival. Fifty-one patients were available for survival analysis. DNA flow cytometry was performed on 30 patients and compared with PCNA score and survival. The prognostic variables were analyzed with a multivariate technique using the Cox hazard model. RESULTS Nuclear atypia (mild, 14; moderate, 23; severe, 14), mitosis (low, 34; high, 17), and tumor necrosis (< 50%, 37; > 50%, 14) were found to highly affect survival in the log-rank test (P < 0.01). Sixteen patients with a high (> or = 12.5) PCNA score had a worse survival (P < 0.01) than did the 35 patients with a low (< 12.5) PCNA score. In patients in whom DNA flow cytometry was performed, the S+G2M-phase fraction showed no correlation with the clinical outcome. However, there was a significant relationship between the extent of PCNA staining and S+G2M fraction (correlation coefficient [CC] = 0.54; P = 0.002), although the CC between PCNA staining and the mitotic count was only 0.38. However, the ploidy pattern was not related to PCNA scores or prognosis. In a multivariate analysis, a high PCNA score (P = 0.017) and severe nuclear atypia (P = 0.0003) were strong prognostic factors. CONCLUSIONS The results suggest that a high PCNA score is one of the poor prognostic factors in synovial sarcoma.