Transrectal ultrasound (TRUS) has shown some promise in the staging of prostate cancer, while prostate-specific antigen (PSA) alone is insufficient. By combining prospective TRUS evaluation with retrospective PSA analysis, we demonstrated an increased accuracy of this combined staging method over TRUS alone. In 48 men undergoing radical prostatectomy, TRUS was performed and PSA measured pre-operatively. On the basis of TRUS, tumours were classified as contained or uncontained. An "expected" PSA value was then calculated for each patient as follows: K x volume of hypoechoic area +0.07 x prostate volume where K = 2.1 if the combined Gleason score of the initial biopsy was > or = 7, or 4.2 if the score was < or = 6. If a patient's pre-operative PSA value was less than or equal to the expected PSA, his tumour was judged to be contained. Staging by both TRUS and PSA was combined, so that if the tumour was judged uncontained by either parameter, the combined prediction was uncontained. Results of the combined staging were: sensitivity 84%, specificity 82%, positive predictive value 94%, negative predictive value 60%, accuracy 83%. This PSA formula, which takes into account the size and grade of the lesion rather than an arbitrary cut-off value, enhances the local staging of prostate cancer by TRUS.