Advances in genomics offer new strategies for assessing the association of common genetic variations at multiple loci and risk of many diseases, including colorectal cancer. Low-penetrance alleles of genes in many biological pathways, such as DNA repair, metabolism, inflammation, cell cycle, apoptosis, and Wnt signaling, may influence the risk of nonfamilial colorectal cancer. To identify susceptibility genes for colorectal cancer, we designed a large-scale case-control association study nested within the Nurses' Health Study (190 cases and 190 controls) and the Health Professionals' Follow-up Study (168 cases and 168 controls). We used a custom GoldenGate (Illumina) oligonucleotide pool assay including 1,536 single nucleotide polymorphisms (SNP) selected in candidate genes from cancer-related pathways, which have been sequenced and genotyped in the SNP500Cancer project; 1,412 of the 1,536 (92%) of the SNPs were genotyped successfully within 388 genes. SNPs in high linkage disequilibrium (r(2) >/= 0.90) with another assayed SNP were excluded from further analyses. As expected by chance (and not significant compared with a corrected Bonferroni P = 0.00004), in the additive model, 11 of 1,253 (0.9%) SNPs had a P(trend) < 0.01 and 38 of 1,253 (3.0%) SNPs had a P(trend) >/= 0.01 and P(trend) < 0.05. Of note, the MGMT Lys(178)Arg (rs2308237) SNP, in linkage disequilibrium with the previously reported MGMT Ile(143)Val SNP, had an inverse association with colorectal cancer risk (MGMT Lys(178)Arg: odds ratio, 0.52; 95% confidence interval, 0.35-0.78; unadjusted P(trend) = 0.0003 for the additive model; gene-based test global P = 0.00003). The SNP500Cancer database and the Illumina GoldenGate Assay allowed us to test a larger number of SNPs than previously possible. We identified several SNPs worthy of investigation in larger studies.