Preoperative transrectal ultrasonograms in 70 patients who underwent radical prostatectomy were compared retrospectively to the histological features of whole mount sections of the surgical specimens. In 42 cases (60 per cent) the tumor could be appreciated as a hypoechoic area on the sonogram. In only 1 case did the tumor appear as a hyperechoic area. In the other 27 cases (39 per cent) the tumor was isoechoic and could not be distinguished clearly from the surrounding tissue. The smallest tumor visualized sonographically measured 4.4 mm. in diameter in the surgical specimen. There was a statistically significant correlation among tumor echogenicity, the actual tumor size and the Gleason grade. Although there was a linear correlation between the size (maximum diameter) of the tumor measured by ultrasound and that determined from the histological sections, ultrasonography generally underestimated the size of the tumor, with the maximum diameter measured sonographically being approximately 4.8 mm. smaller than the diameter measured in the whole mount sections. Of 25 nonpalpable (stage A) tumors 9 (36 per cent) were visualized on ultrasound, while of 45 palpable (stage B) tumors 11 (21 per cent) were not. Although most clinically recognized tumors can be appreciated sonographically, the tumors that are visualized tend to be larger, less well differentiated and palpable. These features favor the use of sonography to stage and monitor established prostate cancers, and suggest that tumors detected by sonography in an early detection program are likely to be clinically important cancers.