A series of 145 patients with T1-2NOMO prostatic carcinoma underwent radical prostatectomy. Specimens were analysed by Gleason sum and categorised according to whether the disease was organ-confined, specimen-confined or margin-positive. The failure of treatment was indicated by elevation of acid phosphatase, by the appearance of biopsy proven local disease or distant nodal disease, or by parenchymal or nodal disease by any imaging modality. Analysis of the data demonstrated that the outcome of surgical intervention was related to the anatomical extent of disease. The Gleason sum was related to the extent of disease and predicted the probability of disease outside the organ of origin.