We assessed the use of clinical stage, serum prostate specific antigen, DNA ploidy, proliferation, and traditional histologic findings from the biopsy to predict prostate cancer progression after radical retropubic prostatectomy. Between 1995 and 1998, 454 consecutive patients with cancer on biopsy were treated by radical retropubic prostatectomy. Preoperative serum prostate specific antigen, clinical stage, Gleason score, percentage of cores and surface area positive for cancer, perineural invasion, and DNA ploidy and MIB-1 immunostain quantitation by image analysis were evaluated in a multivariate Cox proportional hazards regression model to predict cancer progression. Cancer progression was defined as a postoperative serum prostate specific antigen level of > or = 0.4 ng/mL, local recurrence, or systemic progression. Mean follow-up was 3.4 years (range 17 days to 5.8 years). Cancer progression was observed in 73 patients with a mean time to progression of 2.1 years (range 33 days to 5.1 years). Gleason score (p <0.001), MIB-1 cancer proliferation (p = 0.008), and perineural invasion (p = 0.008) were significantly associated with progression. Patients with cancer Gleason scores of 7 and >7 had a 2.5-fold and nearly 4-fold increased risk, respectively, of cancer progression compared with patients with cancer Gleason scores of < or = 6. Patients with perineural invasion at biopsy were twice as likely to progress compared with patients without perineural invasion. Each 1-unit increase in MIB-1 on the natural logarithmic scale increased the risk of cancer progression by 64%. Cancer progression models that include serum prostate specific antigen and clinical stage may require revision to incorporate perineural invasion and MIB-1 proliferative activity in addition to Gleason score.