Although the preoperative serum prostate-specific antigen (PSA) concentration has been demonstrated to be directly proportional to pathologic stage, it has been found to be unreliable for predicting final pathologic stage on an individual basis. Statistical analysis was thus performed to determine whether combining various preoperative parameters could enhance the predictive power of PSA. Multivariate logistic regression analysis revealed that, indeed, local clinical stage (determined by digital rectal examination) and tumor grade (determined from the biopsy specimen) significantly enhanced the predictive power of PSA (P < 0.0001 and < 0.0001, respectively). These preoperative parameters were thus all combined into a model from which probability plots were generated. These probability plots allow the practicing urologist to estimate preoperatively for an individual patient the probability of a given pathologic stage and DNA ploidy status. For example, the predicted probability of organ-confined disease for a patient with a PSA value of 10 ng/mL, when used alone, is 53 percent. However, the predicted probability of organ-confined disease for a patient with a PSA value of 10 ng/mL, clinical Stage B2, and high-tumor grade is only 30 percent. Thus, the findings of this investigation allow the physician maximal use of the preoperative variables to discuss more accurately with the patient what might be expected at the time of operation and to plan more precisely the surgical procedure that will possibly ensure complete removal of the prostate cancer with the least compromise to normal physiologic function (continence and potency).