A total of 51 patients underwent transrectal ultrasound of the prostate before radical cystoprostatectomy for transitional cell carcinoma of the bladder. Each had a normal prostate by digital rectal examination, no history of prostatic adenocarcinoma and no invasion of the prostate by transitional cell carcinoma. Real-time and step-sectioned ultrasound images were interpreted at the time of sonography and results were compared to pathological examination of the step-sectioned prostate specimen. When adenocarcinoma was identified in the specimen, cancer volume was determined. Positive ultrasound scans consisted of those exhibiting hypoechoic lesions. Hypoechogenicity due to transurethral resection defects, benign hyperplasia, vascular structures or shadowing from calcifications was not considered positive. Of 51 patients 27 (52.9%) exhibited no abnormality on ultrasound and were free of cancer in the prostate specimen, while 8 (15.7%) demonstrated a hypoechoic lesion that was proved to be prostate cancer. Seven patients (13.7%) with normal transrectal ultrasound scans had adenocarcinoma of the prostate, while 9 (17.6%) had lesions on ultrasound but no cancer. Based on these results, transrectal ultrasound has a sensitivity of 53.3% and a specificity of 75%. Further analysis reveals that transrectal ultrasound is more accurate in the detection of cancers of greater than 0.20 cc in volume than those of 0.20 cc or less. Transrectal ultrasound also is more accurate in the detection of peripheral zone than transition zone cancers.