In a retrospective correlative analysis of postmortem findings and antemortem transurethral resection specimens from 81 patients with prostatic carcinoma a simple and reliable prediction index was found. This index was derived from the transurethral resection specimen in regard to the probability of ultimate bone metastasis by tumor and was obtained by dividing the number of prostatic tissue chips involved with tumor by the total number of prostatic tissue chips examined microscopically. We found that when 75 per cent or more of the transurethral resection chips were involved with tumor, regardless of the degree of tumor differentiation, the probability of bone metastasis was more than 9.0 (greater than 90 per cent), whereas the probability was 0.58 for 50 to 75 per cent involvement, 0.38 for 30 to 50 per cent involvement, 0.27 for 20 to 30 per cent involvement, 0.10 for 10 to 20 per cent involvement and 0.06 for less than 10 per cent involvement. Other tumor grading systems, such as the Gleason system, also were compared. A good correlation was obtained by the Gleason grading system in the prediction of bone metastasis but the system is not as simple or as reproducible as the tumor involvement index described herein. Other parameters, such as prostatic acid phosphatase and prostatic specific antigen as determined by the immunoperoxidase method, had no specific or significant value in the prediction of bone metastasis. Our findings reaffirm the relatively old concept that the extensiveness of tumor involvement in the transurethral resection specimen is the single most important factor in the prediction of the clinical behavior of the prostatic carcinoma.