We examined whether selected polymorphisms in 11 candidate genes and serum levels of insulin-like growth factor I (IGF-I) help predict the presence of prostate cancer among patients prescreened with prostate-specific antigen (PSA) and digital rectal exam (DRE). We studied 1031 consecutive men who underwent one or more prostate biopsies because of an elevated PSA level (>4 ng/ml) or an abnormal DRE. Eleven candidate genes were examined, including the androgen receptor, SRD5A2, CYP17, CYP3A4, vitamin D receptor, PSA, GST-T1, GST-M1, GST-P1, IGF-I, and IGF binding protein 3. We also measured serum IGF-I levels before biopsy. Of the 1031 men, 483 had cancer on any biopsy (cases) and 548 men had no cancer (controls). Age, ethnicity, total PSA, and DRE result were strongly predictive of the presence of prostate cancer. The mean IGF-I level for cases (119.4 ng/ml) was lower than for controls (124.4 ng/ml, P = 0.05) and were not predictive for the presence of prostate cancer. We found no associations between the androgen receptor, SRD5A2, CYP17, CYP3A4, vitamin D receptor, GST-M1, GST-P1, and IGF binding protein 3 genotypes and prostate cancer risk. The adjusted odds ratios for having prostate cancer for patients with the GST-T1 and IGF-I variant alleles were 1.64 (95% confidence interval, 1.1-2.4; P = 0.01) and 1.70 (95% confidence interval, 1.1-2.7; P = 0.02), respectively. Nine of 11 candidate genes were not significantly predictive for prostate cancer in a clinical setting. The GST-T1 and IGF-I polymorphisms demonstrated modest associations with prostate cancer risk. IGF-I levels were not helpful in identifying patients with prostate cancer at the time of biopsy.