Mucosal melanoma (MM) is a highly vascularized tumor with an extremely poor prognosis. In this randomized, open-label, phase II study, we characterized the efficacy and safety of bevacizumab in combination with carboplatin plus paclitaxel (CPB) in patients with previously untreated advanced MM.Patients were randomly assigned in a 2:1 ratio to receive carboplatin (area under the curve, 5) plus paclitaxel (175 mg/m2) once every 4 weeks in combination with (CPB arm, 5 mg/kg) or without (CP arm) bevacizumab once every 2 weeks. Progression-free survival (PFS) was the primary end point. Secondary end points included overall survival (OS), objective response rate, and adverse events.We recruited 114 patients to our study. The median PFS was significantly longer in the CPB arm (4.8 months; 95% CI, 3.6 to 6.0 months) than in the CP arm (3.0 months; 95% CI, 1.7 to 4.3 months) (hazard ratio, 0.461; 95% CI, 0.306 to 0.695; P < .001). Objective response rates were 19.7% and 13.2%, respectively (P = .384). The median OS was also significantly longer in the CPB arm than in the CP arm (13.6 v 9.0 months; hazard ratio, 0.611; 95% CI, 0.407 to 0.917; P = .017). No new safety signals were observed.PFS and OS were significantly better in patients with metastatic MM who received bevacizumab in addition to CPB than in those who received CPB alone. A phase III study should be performed to confirm these benefits (ClinicalTrials.gov identifier: NCT02023710).