In 1995, there will be 244,000 new cases of prostate cancer, and 40,400 deaths from prostate cancer, among men in the United States. The American Cancer Society reports that the incidence rate of prostate cancer is increasing at an accelerated pace, and was 21 percent higher in 1994 than in 1993. The major reason for this steep rise is likely to be due to increased popularity of prostate cancer screening which, by identifying latent, asymptomatic cases, may convert them into clinical cases. Is screening--an important means of cancer control for many sites--a reasonable approach for prostate cancer control? The answer is not straightforward because prostate cancer is not one, but three diseases: a latent form which will cause no harm; a progressive form which will become symptomatic and can kill; and a rapidly progressive form so malignant that it is likely to kill, whether detected early or late. Screen-detection may be worthwhile only for the second form, as tumors of the first form need never be detected, and tumors of the third form progress so rapidly that timely screen-detection is nearly impossible and, if accomplished, may be valueless. As there is no way to differentiate among the three diseases when screening, the possible deleterious effects of screen-detection must be weighed against the benefits.