Genitourinary cancers--bladders, testis and prostate--account for almost 50,000 deaths per year. Epidemiologic data suggest that individuals with low serum retinoid levels or low dietary intake of retinoid-containing foodstuffs have an increased risk of bladder and prostate cancer. Preclinical investigations show that a variety of retinoids suppress the proliferation of prostate and bladder cancer cells and induce differentiation in teratocarcinoma cells. Retinoids prevent the emergence of murine bladder and prostate cancers in carcinogen-treated animals. Clinical data are disappointing or inconclusive. A single well-conducted phase 2 trial of 13-cis retinoic acid in patients with germ cell tumors was negative. Several historically controlled as well as prospectively randomized, placebo-controlled trials of retinoids in superficial bladder cancer have failed to provide evidence of the efficacy of retinoids. Two studies of all-trans-retinoic acid in advanced prostate cancer have been negative. Despite compelling preclinical rationale, retinoids have failed to yield positive results in the clinical management of prostate, bladder or testis cancers. Further work is needed to define subsets of patients in whom retinoids might be active, and whether new retinoids or new approaches to retinoid delivery will improve the clinical usefulness of retinoids in these tumors.