Insulin-like growth factors (IGFs) may play a role in prostate growth, hyperplasia, and malignancy. High plasma IGF-I has been associated with increased prostate cancer risk. In a prospective, cohort, case-control study in the Baltimore Longitudinal Study on Aging population, we examined prostate volume by magnetic resonance imaging, and prostate-specific antigen (PSA), IGF-I, IGF-II, and IGF-binding protein-3 (IGFBP-3) in sera obtained approximately 9 yr before diagnosis of prostate cancer in cases (n = 72) or age-matched controls (n = 127) and in 76 additional Baltimore Longitudinal Study on Aging men (normal subjects) with measured prostate volumes and no prostate cancer. We calculated adjusted odds ratios (OR) by logistic regression, relative risks for significant ORs, and receiver operator curves for prostate cancer, using serum measures alone and in combination. Adjusted ORs for the high vs. low tertile were: for IGF-I, 3.1 [confidence interval (CI), 1.1-8.7]; for IGF-II, 0.2 (CI, 0.07-0.6); for IGFBP-3, 0.71 (CI, 0.3-1.7); and for PSA, 12.5 (CI, 3.8-40.9). For significant ORs, relative risk estimates remained significant at 2.0 for IGF-I, 0.3 for IGF-II, and 5.5 for PSA. Receiver operator curves showed PSA to be the most powerful predictor of prostate cancer. Adding IGF-II to PSA improved prediction. IGF-II was significantly and inversely related (r = -0.219; P < 0.01) and PSA was directly and significantly related (r = 0.461; P < 0.0001) to prostate volume, whereas IGF-I and IBFBP-3 were not. High IGF-I and low IGF-II are independently associated with increased risk of prostate cancer, but PSA level is a much stronger predictor of prostate cancer in the ensuing 10 yr than either IGF-I or IGF-II. The absence of a relationship of IGF-I to prostate size is inconsistent with increased ascertainment in men with large prostates as the source of greater prostate cancer risk associated with IGF-I. Our data suggest that IGF-II may inhibit both prostate growth and development of prostate cancer.