A retrospective study of five years' experience with fourth-generation computerized tomography (CT) scan was undertaken to assess the frequency of understaging in prostate cancer. A total of 160 patients with preoperative scans were surgically staged. In 10 patients, the operation was aborted after pelvic node dissection had revealed unsuspected metastatic involvement. Based on the histopathologic evidence of local tumor invasion, extension into seminal vesicles or pelvic lymph nodes, restaging was required in 78 percent of cases. Accuracy was 24 percent for capsular extension, 69 percent for seminal vesicle invasion, and 72 percent for lymphadenopathy. The poor yield of CT scan as a preoperative staging modality is demonstrated. Recent advances in the understanding and management of prostatic cancer require reassessing patient benefit and cost effectiveness of available imaging techniques, focusing on the problem of detecting nodal metastases, and predicting tumor spread to regional lymph nodes by accurately evaluating the primary neoplasm. We conclude that CT scan fails to demonstrate the required precision needed to evaluate local tumor spread; therefore, this goal must be pursued with newer imaging modalities.