Background: There is an unmet need for predictive markers for the antiangiogenic agent bevacizumab in metastatic colorectal cancer (mCRC). We aimed to assess whether the location of the primary tumor is associated with bevacizumab effectiveness when combined with capecitabine and oxaliplatin (CAPEOX) in the first-line treatment of patients with mCRC. Patients and methods: A cohort of 667 consecutive patients with mCRC from the general community treated from 2006 to 2011 with CAPEOX and bevacizumab as standard first-line therapy was compared with a cohort of 213 patients treated with CAPEOX from2003 to 2006, before bevacizumab was approved. Main outcome measures were progression-free survival (PFS) and overall survival (OS). Differences in outcome were tested using Kaplan-Meier curves and log-rank tests, and multivariate analyses were carried out using Cox Proportional Hazards models. Results: Patients treated with CAPEOX and bevacizumab with primary tumors originating in the sigmoid colon and rectum had a significantly better outcome than patients with primary tumors originating from the cecum to the descending colon, both for PFS (median PFS 9.3 versus 7.2 months; hazard ratio (HR) 0.68,95% confidence interval (CI) 0.56-0.82) and for OS (median OS 23.5 versus 13.0 months; HR 0.47, 95% CI 0.38-0.57). This difference was confirmed in multivariate analyses after adjustment for other potentially prognostic factors. For patients treated with CAPEOX, there was no association between primarytumor location and outcome, neither in unadjusted nor adjusted analyses. Conclusions: The addition ofbevacizumab to CAPEOX in first-line treatment of patients with mCRC may primarily benefit patients with primary tumors originating in the rectum and sigmoid colon. This hypothesis needs to be validated in data from completed randomized trials. ClinicalTrials.gov identification number: NCT00212615.