PURPOSE Interventional hepatic arterial infusion chemotherapy of infusional fluorouracil, leucovorin, and oxaliplatin (HAIC-FO) displayed an encouraging safety profile and antitumor activity in a previous phase II trial and a propensity-score-matching study involving patients with locally advanced hepatocellular carcinoma (HCC). METHODS In this open-label, phase III trial, patients with advanced HCC, previously untreated with systemic therapy, were randomly assigned in a 1:1 ratio to receive HAIC-FO or sorafenib. The primary end point was overall survival (OS) in the intention-to-treat population. An exploratory model for predicting the efficacy of HAIC-FO on the basis of genomic sequencing was developed. RESULTS Between May 2017 and May 2020, 262 patients were randomly assigned. The median tumor size was 11.2 cm (interquartile range, 8.5-13.7 cm). Macrovascular invasion was present in 65.6%, and the percentage of patients with > 50% tumor volume involvement of the liver and/or Vp-4 portal vein tumor thrombosis was 49.2%. At data cutoff (October 31, 2020), median OS was 13.9 months for HAIC-FO and 8.2 for sorafenib (hazard ratio [HR] 0.408; 95% CI, 0.301 to 0.552; P < .001). Tumor downstaging occurred in 16 (12.3% of 130) patients receiving HAIC-FO, including 15 receiving curative surgery or ablation, and finally achieving a median OS of 20.8 months, with a 1-year OS rate of 93.8%. In high-risk subpopulations, OS was significantly longer with HAIC-FO than with sorafenib (10.8 months v 5.7 months; HR 0.343; 95% CI, 0.219 to 0.538; P < .001). A newly developed 15-mutant-gene prediction model identified 83% of patients with response to HAIC-FO. HAIC-FO responders had longer OS than HAIC-FO nonresponders (19.3 months v 10.6 months; HR 0.323; 95% CI, 0.186 to 0.560; P = .002). CONCLUSION HAIC-FO achieved better survival outcomes than sorafenib in advanced HCC, even in association with a high intrahepatic disease burden.