Background: Cancers of the colon and rectum are largely considered to be molecularly comparable diseases, harboring activating mutations in KRAS, BRAF, NRAS, and PIK3CA. These mutations have an impact on both prognosis and therapeutic options. In this study, we used a multiplexed clinical assay (SNaPshot) (Dias-Santagata D, EMBO Mol Med. 2010 May;2(5):146-58) performed on nucleic acid derived from formalin-fixed tissue which tests for 120 previously described mutations across 13 cancer genes and examined the relationship with clinical features and disease site. Methods: 185 consecutive metastatic colorectal patients were identified in our clinic and underwent testing with SNaPshot in an IRB approved protocol. Mutational analysis was performed across 13 genes, including KRAS codons 12 and 13, NRAS codons 12, 13, and 61, BRAF codon 600, and PIK3CA codons 88, 542, 545-546, 1047, and 1049. Patient charts were reviewed for demographic data, primary site, metastatic sites at presentation and last follow-up. Fisher’s exact test was performed to determine statistical significance. Results: Of the 185 patients, 51 (38%) had a rectal primary and 134 (62%) had a colon primary. Median age was 55 years (29-83). There were 78 females (42%). Comparative mutation rate for rectal vs colon primary (respectively) was: KRAS -33% vs 36% (NS), BRAF – 2% vs 11% (P=0.07), PIK3CA - 12% vs 13% (NS), dual PIK3CA/KRAS- 8% vs 5% (NS), and NRAS - 14% vs 1% (P=0.001). The single colon patient with an NRAS mutation had a sigmoid colon primary. Rectal primary was associated with a higher rate of lung metastases at presentation (43%) than colon (21%), P=0.003. NRAS mutation was associated with an 88% (7/8) rate of lung metastases and a 0% rate of bone and CNS metastases. Conclusions: Rectal and colon patients have similar rates of KRAS and PIK3CA mutations. However, BRAF mutations are more common in colon cancer. NRAS mutations are exclusively found in rectosigmoid cancers and may have a different biology than other colorectal cancers. These data suggest that primary tumor location may provide a means to enrich a population for a genotype-directed study.