Summary Background Bevacizumab improves the efficacy of oxaliplatin-based chemotherapy in metastatic colorectal cancer. Our aim was to assess the use of bevacizumab in combination with oxaliplatin-based chemotherapy in the adjuvant treatment of patients with resected stage III or high-risk stage II colon carcinoma. Methods Patients from 330 centres in 34 countries were enrolled into this phase 3, open-label randomised trial. Patients with curatively resected stage III or high-risk stage II colon carcinoma were randomly assigned (1:1:1) to receive FOLFOX4 (oxaliplatin 85 mg/m2 , leucovorin 200 mg/m2 , and fluorouracil 400 mg/m2 bolus plus 600 mg/m2 22-h continuous infusion on day 1; leucovorin 200 mg/m2 plus fluorouracil 400 mg/m2 bolus plus 600 mg/m2 22-h continuous infusion on day 2) every 2 weeks for 12 cycles; bevacizumab 5 mg/kg plus FOLFOX4 (every 2 weeks for 12 cycles) followed by bevacizumab monotherapy 7·5 mg/kg every 3 weeks (eight cycles over 24 weeks); or bevacizumab 7·5 mg/kg plus XELOX (oxaliplatin 130 mg/m2 on day 1 every 2 weeks plus oral capecitabine 1000 mg/m2 twice daily on days 1–15) every 3 weeks for eight cycles followed by bevacizumab monotherapy 7·5 mg/kg every 3 weeks (eight cycles over 24 weeks). Block randomisation was done with a central interactive computerised system, stratified by geographic region and disease stage. Surgery with curative intent occurred 4–8 weeks before randomisation. The primary endpoint was disease-free survival, analysed for all randomised patients with stage III disease. This study is registered with ClinicalTrials.gov , number NCT00112918. Findings Of the total intention-to-treat population (n=3451), 2867 patients had stage III disease, of whom 955 were randomly assigned to receive FOLFOX4, 960 to receive bevacizumab–FOLFOX4, and 952 to receive bevacizumab–XELOX. After a median follow-up of 48 months (range 0–66 months), 237 patients (25%) in the FOLFOX4 group, 280 (29%) in the bevacizumab–FOLFOX4 group, and 253 (27%) in the bevacizumab–XELOX group had relapsed, developed a new colon cancer, or died. The disease-free survival hazard ratio for bevacizumab–FOLFOX4 versus FOLFOX4 was 1·17 (95% CI 0·98–1·39; p=0·07), and for bevacizumab–XELOX versus FOLFOX4 was 1·07 (0·90–1·28; p=0·44). After a minimum follow-up of 60 months, the overall survival hazard ratio for bevacizumab–FOLFOX4 versus FOLFOX4 was 1·27 (1·03–1·57; p=0·02), and for bevacizumab–XELOX versus FOLFOX4 was 1·15 (0·93–1·42; p=0·21). The 573 patients with high-risk stage II cancer were included in the safety analysis. The most common grade 3–5 adverse events were neutropenia (FOLFOX4: 477 42% of 1126 patients, bevacizumab-FOLFOX4: 416 36% of 1145 patients, and bevacizumab–XELOX: 74 7% of 1135 patients), diarrhoea (110 10%, 135 12%, and 181 16%, respectively), and hypertension (12 1%, 122 11%, and 116 10%, respectively). Serious adverse events were more common in the bevacizumab groups (bevacizumab–FOLFOX4: 297 26%; bevacizumab–XELOX: 284 25%) than in the FOLFOX4 group (226 20%). Treatment-related deaths were reported in one patient receiving FOLFOX4, two receiving bevacizumab–FOLFOX4, and five receiving bevacizumab–XELOX. Interpretation Bevacizumab does not prolong disease-free survival when added to adjuvant chemotherapy in resected stage III colon cancer. Overall survival data suggest a potential detrimental effect with bevacizumab plus oxaliplatin-based adjuvant therapy in these patients. On the basis of these and other data, we do not recommend the use of bevacizumab in the adjuvant treatment of patients with curatively resected stage III colon cancer. Funding Genentech, Roche, and Chugai.