The human androgen receptor gene contains polymorphic CAG and GGC repeats in exon 1. We investigated whether the number of CAG and/or GGC repeats is related to prostate cancer risk in a case-control study nested within the beta Carotene and Retinol Efficacy Trial. Among 300 cases and 300 controls, we did not observe any increase in risk associated with fewer CAG or GGC repeats. We observed a nonsignificant decrease in risk associated with each unit of decrease in CAG length [odds ratio (OR), 0.98; 95% confidence interval (CI), 0.93-1.03). Men with CAG <22 had a relative risk of prostate cancer of 0.89 (95% CI, 0.65-1.23) compared with men with CAG > or =22. There was no appreciable difference in the mean number of GGC repeats between cases and controls; the estimated change in the risk of prostate cancer associated with one fewer GGC repeat was 0.97 (95% CI, 0.88-1.06). The risk in men at or below the mean number of GGC repeats (17) was 0.80 (95% CI, 0.57-1.12). In contrast to prior reports, men with both short CAG (<22) and short GGC (< or =17) repeats were not at increased risk of prostate cancer (OR, 0.56; 95% CI, 0.32-0.98), compared with men with > or =22 CAG repeats and >17 GGC repeats. Our results do not support the hypothesis that a small number of CAG or GGC repeats in the androgen receptor gene increases a man's risk of prostate cancer.