The relation between serum vitamin A measurements made at baseline examination (1971-1975) and subsequent development of prostate cancer was examined in the National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study (1981-1984). The analytic cohort consisted of 2440 men 50 years of age or older who were followed for a median of 10 years. A total of 84 men developed prostate cancer. The mean level of serum vitamin A was significantly lower (P less than 0.01) in prostate cancer cases than in noncases. Considered as a continuous variable or in quartiles, a statistically significant (P less than 0.005 or P less than 0.02, respectively) trend was observed for increased risk of prostate cancer with decreasing levels of serum vitamin A. Adjusted for age and race, men in the lowest quartile had a relative risk of 2.2 (95% confidence intervals, 1.1, 4.3) compared to those in the highest quartile. The elevated risk of prostate cancer associated with the lowest quartile of serum vitamin A levels did not attenuate with increasing time between blood drawing and diagnosis, suggesting that metabolic effects of early disease are an unlikely explanation of these results. The inverse association between serum vitamin A and prostate cancer incidence was independent of age at examination and several other possible confounding variables. This is the first prospective study of serum vitamin A and prostate cancer to include a large (84) number of cases.