Screening for prostate cancer represents a clinical dilemma with no clear evidence to suggest decreased mortality from any diagnostic test. We now possess new knowledge regarding optimal combinations of DRE, TRUS, and PSA. While DRE and TRUS may be too subjective and PSA too nonspecific, their combined predictive values identify not only men at high risk but also those for whom continued frequent screening may not be cost effective. A monoclonal PSA decision level of no more than 4.0 ng/ml should be used, since 40 percent of cancers detected from 4.0 to 10.0 ng/ml already have extracapsular extension. Assuming that DRE is performed by experienced examiners, the combination of PSA and DRE should produce cost-effective early detection and minimize missed cancers below 4.0 ng/ml. TRUS should be reserved for those patients with either PSA elevations and/or DRE abnormalities. The use of TRUS gland volume data to further modify PSA decision levels, such as the "predicted" PSA concept, may also improve TRUS biopsy criteria and predictive values. Prostate cancer detection can then be objectively limited to a small percentage of the population and better selected for earlier, more localized disease. The ultimate decrease in mortality from screening remains to be demonstrated in randomized trials or observed only after decades of increased public awareness about prompt early detection combined with effective, definitive therapy.