We evaluated serum prostate specific antigen before and after radical prostatectomy. In 100 consecutive patients who underwent radical prostatectomy, preoperative prostate specific antigen levels tended to increase with the increasing severity of pathological stage. However, even at levels of greater than 10 ng. per ml. the positive and negative predictive values (78 and 61 per cent, respectively) of prostate specific antigen to predict extracapsular disease were not sufficient to make this test useful alone for staging. In theory, after radical prostatectomy prostate specific antigen should be zero if no remaining prostatic tissue is present. Tests of precision and analytical sensitivity in our laboratory using a commercial prostate specific antigen assay revealed that a value of 0.4 ng. per ml. or more is different from zero at a greater than 95 per cent confidence level. With this guideline we evaluated the meaning of prostate specific antigen levels 3 to 6 months after radical prostatectomy in 59 men. Among men whose prostate specific antigen level was less than 0.4 ng. per ml. only 9 per cent demonstrated recurrence as evidenced by the development of positive bone scan or progressively elevated prostate specific antigen levels within 6 to 50 months. Alternatively, in men whose 3 to 6-month prostate specific antigen level was 0.4 ng per ml. or more there was evidence of recurrence in 100 per cent within 6 to 49 months (p less than 0.0001). Progressively elevated (more than 0.4 ng. per ml.) prostate specific antigen levels preceded recurrence from 12 to 43 months in all 6 patients who had positive bone scans, while increasing prostate specific antigen levels since radical prostatectomy have continued from 9 to 65 months in the 11 patients who have no radiological evidence of recurrent disease to date. Prostatic acid phosphatase serum values after radical prostatectomy were not useful to predict persistent disease. Prostate specific antigen values 3 to 6 months after radical prostatectomy are a sensitive indicator of persistent disease after radical prostatectomy and often precede other evidence of this occurrence by many years. This fact may alter concepts about surgical results, and possibly shorten and sharpen clinical studies involving adjuvant therapy after radical prostatectomy.