Recent epidemiological studies suggest an association between higher blood levels of insulin-like growth factor I (IGF-I) and increased risk of prostate cancer. We evaluated the association between prediagnostic levels of IGF-I and insulin-like growth factor binding protein 3 (IGFBP-3) and prostate cancer risk in a nested case-control study within the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study. Within the same cohort (using different cases and controls who had sequential serum samples available) we also examined changes in serum IGF-I and IGFBP-3 levels over time by case status. The risk association study included incident prostate cancer cases (n = 100) diagnosed at least 5 years after baseline blood draw (range, 5-12 years; median 9 years) and frequency-matched (4:1) controls. The sequential serum study included all of the prostate cancer cases (n = 21) with prediagnostic (2-3 years before diagnosis) and diagnostic serum available, and pair-matched controls (1:1). An ELISA was used to quantitate serum levels of IGF-I and IGFBP-3 for both studies. The association between IGF-I or IGFBP-3 and prostate cancer risk was assessed using conditional logistic regression, and paired t tests were used to evaluate case-control differences in change in serum analytes over time. We found no significant association between either IGF-I or IGFBP-3 and prostate cancer risk. In a multivariate analysis, we observed an odds ratio of 0.52 (95% confidence interval, 0.23-1.16) for the fourth versus the first quartile of serum IGF-I. Serum IGF-I, but not IGFBP-3, increased significantly over time in cases (18% increase) but not controls (4% decrease; P = 0.02). In contrast to previous reports, we found no evidence to support a causal association between serum IGF-I or IGFBP-3 and the risk of prostate cancer. It is possible that serum IGF-I may be serving as a tumor marker rather than an etiologic factor in prostate cancer.