Although tumor volume is an important factor in predicting prognosis in carcinoma of the prostate, direct and accurate estimation of tumor volume is not practical clinically at present because the tumor may not always be palpable (stage A) and when palpable it is difficult to estimate volume in 3 dimensions. For this reason the clinical staging of prostate cancer currently is based on estimations of the per cent of gland involved with tumor: in stage A by per cent of tissue involved with cancer and in stage B by digital palpation (less than 1 lobe, 1 lobe and 2 lobes). In stage A prostate cancer the per cent of the specimen involved with tumor and the volume of tumor have been shown to correlate with tumor progression. Our study was designed to determine if either or both of these morphometric factors would be good predictors of pathological stage in stage B prostate cancer. We analyzed 56 step-sectioned radical prostatectomy specimens: 28 without capsular penetration, 15 with capsular penetration only and 13 with seminal vesicle involvement. The per cent of gland involved with tumor (correlation coefficient 0.67, p less than 0.001) and tumor volume (correlation coefficient 0.55, p less than 0.001) correlated well with pathological stage. Stepwise linear regression showed that the combination of the per cent of gland involved with tumor and the total Gleason grade was statistically the best predictor of pathological stage.