PURPOSE About half of patients with metastatic uveal melanoma present with isolated liver metastasis, in whom the median survival is 6-12 months. The few systemic treatment options available only moderately prolong survival. Isolated hepatic perfusion (IHP) with melphalan is a regional treatment option, but prospective efficacy and safety data are lacking. METHODS In this multicenter, randomized, open-label, phase III trial, patients with previously untreated isolated liver metastases from uveal melanoma were randomly assigned to receive a one-time treatment with IHP with melphalan or best alternative care (control group). The primary end point was overall survival at 24 months. Here, we report the secondary outcomes of response according to RECIST 1.1 criteria, progression-free survival (PFS), hepatic PFS (hPFS), and safety. RESULTS Ninety-three patients were randomly assigned, and 87 patients were assigned to either IHP (n = 43) or a control group receiving the investigator's choice of treatment (n = 44). In the control group, 49% received chemotherapy, 39% immune checkpoint inhibitors, and 9% locoregional treatment other than IHP. In an intention-to-treat analysis, the overall response rates (ORRs) were 40% versus 4.5% in the IHP and control groups, respectively ( P < .0001). The median PFS was 7.4 months versus 3.3 months ( P < .0001), with a hazard ratio of 0.21 (95% CI, 0.12 to 0.36), and the median hPFS was 9.1 months versus 3.3 months ( P < .0001), both favoring the IHP arm. There were 11 treatment-related serious adverse events in the IHP group compared with seven in the control group. There was one treatment-related death in the IHP group. CONCLUSION IHP treatment resulted in superior ORR, hPFS, and PFS compared with best alternative care in previously untreated patients with isolated liver metastases from primary uveal melanoma.