Background: The triplet FOLFOXIRI (fluorouracil, leucovorin, oxaliplatin, and irinotecan) plus bevacizumab showed improved outcomes for patients with metastatic colorectal cancer, compared with FOLFIRI (fluorouracil, leucovorin, and irinotecan) plus bevacizumab. However, the actual benefit of the upfront exposure to the three cytotoxic drugs compared with a preplanned sequential strategy of doublets was not clear, and neither was the feasibility or efficacy of therapies after disease progression. We aimed to compare a preplanned strategy of upfront FOLFOXIRI followed by the reintroduction of the same regimen after disease progression versus a sequence of mFOLFOX6 (fluorouracil, leucovorin, and oxaliplatin) and FOLFIRI doublets, in combination with bevacizumab. Methods: TRIBE2 was an open-label, phase 3, randomised study of patients aged 18–75 years with an Eastern Cooperative Oncology Group (ECOG) performance status of 2, with unresectable, previously untreated metastatic colorectal cancer, recruited from 58 Italian oncology units. Patients were stratified according to centre, ECOG performance status, primary tumour location, and previous adjuvant chemotherapy. A randomisation system incorporating a minimisation algorithm was used to randomly assign patients (1:1) via a masked web-based allocation procedure to two different treatment strategies. In the control group, patients received first-line mFOLFOX6 (85 mg/m2 of intravenous oxaliplatin concurrently with 200 mg/m2 of leucovorin over 120 min; 400 mg/m2 intravenous bolus of fluorouracil; 2400 mg/m2 continuous infusion of fluorouracil for 48 h) plus bevacizumab (5 mg/kg intravenously over 30 min) followed by FOLFIRI (180 mg/m2 of intravenous irinotecan over 120 min concurrently with 200 mg/m2 of leucovorin; 400 mg/m2 intravenous bolus of fluorouracil; 2400 mg/m2 continuous infusion of fluorouracil for 48 h) plus bevacizumab after disease progression. In the experimental group, patients received FOLFOXIRI (165 mg/m2 of intravenous irinotecan over 60 min; 85 mg/m2 intravenous oxaliplatin concurrently with 200 mg/m2 of leucovorin over 120 min; 3200 mg/m2 continuous infusion of fluorouracil for 48 h) plus bevacizumab followed by the reintroduction of the same regimen after disease progression. Combination treatments were repeated every 14 days for up to eight cycles followed by fluorouracil and leucovorin (at the same dose administered at the last induction cycle) plus bevacizumab maintenance until disease progression, unacceptable adverse events, or consent withdrawal. Patients and investigators were not masked. The primary endpoint was progression-free survival 2, defined as the time from randomisation to disease progression on any treatment given after first disease progression, or death, analysed by intention to treat. Safety was assessed in patients who received at least one dose of their assigned treatment. Study recruitment is complete and follow-up is ongoing. This trial is registered with Clinicaltrials.gov, NCT02339116. Findings: Between Feb 26, 2015, and May 15, 2017, 679 patients were randomly assigned and received treatment (340 in the control group and 339 in the experimental group). At data cut-off (July 30, 2019) median follow-up was 35·9 months (IQR 30·1–41·4). Median progression-free survival 2 was 19·2 months (95% CI 17·3–21·4) in the experimental group and 16·4 months (15·1–17·5) in the control group (hazard ratio [HR] 0·74, 95% CI 0·63–0·88; p=0·0005). During the first-line treatment, the most frequent of all-cause grade 3–4 events were diarrhoea (57 [17%] vs 18 [5%]), neutropenia (168 [50%] vs 71 [21%]), and arterial hypertension (25 [7%] vs 35 [10%]) in the experimental group compared with the control group. Serious adverse events occurred in 84 (25%) patients in the experimental group and in 56 (17%) patients in the control group. Eight treatment-related deaths were reported in the experimental group (two intestinal occlusions, two intestinal perforations, two sepsis, one myocardial infarction, and one bleeding) and four in the control group (two occlusions, one perforation, and one pulmonary embolism). After first disease progression, no substantial differences in the incidence of grade 3 or 4 adverse events were reported between the control and experimental groups, with the exception of neurotoxicity, which was only reported in the experimental group (six [5%] of 132 patients). Serious adverse events after disease progression occurred in 20 (15%) patients in the experimental group and 25 (12%) in the control group. Three treatment-related deaths after first disease progression were reported in the experimental group (two intestinal occlusions and one sepsis) and four in the control group (one intestinal occlusion, one intestinal perforation, one cerebrovascular event, and one sepsis). Interpretation: Upfront FOLFOXIRI plus bevacizumab followed by the reintroduction of the same regimen after disease progression seems to be a preferable therapeutic strategy to sequential administration of chemotherapy doublets, in combination with bevacizumab, for patients with metastatic colorectal cancer selected according to the study criteria. Funding: The GONO Cooperative Group, the ARCO Foundation, and F Hoffmann–La Roche.