Multifocal prostate cancer has been reported in 50-76% of all cases of radical retropubic prostatectomy (RRP) specimens, but the clinical and prognostic significance of this finding is still unclear. A retrospective analysis of patients who underwent RRP between 1993 and 1997 was performed. Preoperative screening parameters and 4-mm RRP specimen sections were examined. The location, Gleason score, and extracapsular extension of the tumor recorded. Three hundred eight cases were reviewed. Mean follow-up was 4.2 +/- 1 years (range 2-6 years). Two hundred six patients (66.9%) had multifocal prostate cancer and 102 (33.1%) had unifocal prostate cancer. Of those with multifocal disease, 63% had two foci and 37% had three or more foci. There were statistical significant differences between both groups with respect to preoperative prostate-specific antigen (PSA) density of the transition zone (PSA-TZ), free/total (f/t) PSA, as well as percentage of patients with organ confined disease, high-grade tumors, and local recurrence. PSA-TZ (p = .001) and f/t PSA (p = .004) were significantly different between patients with unifocal and multifocal disease (0.9 vs. 2.2 ng/mL/cc and 18% vs. 6.5%, respectively). However, preoperative PSA (11.2 vs. 12.8 ng/mL; p = .09) and PSA density (0.17 vs. 0.19 ng/mL/cc; p = .07) were not able to predict unifocality or multifocality. These data suggest that multifocal prostate cancer is associated with higher grade, stage, and recurrence rate than unifocal prostate cancer. Preoperative PSA-TZ (> 1.5 ng/mL/cc) and f/t PSA (<9%) may predict multifocality in the RRP specimen.