Background: Several recent studies had demonstrated the superiority of Sorafenib/lenvatinib plus transcatheter arterial chemoembolization (TACE) over sorafenib/lenvatinib alone in advanced hepatocellular carcinoma (HCC). Herein, we compared the efficacy and safety between sorafenib plus TACE (SoraTACE) and sorafenib plus hepatic arterial infusion chemotherapy (SoraHAIC) in advanced HCC. Additionally, we validated a predictive model from our previous phase III trial (NCT02973685, cohort1). Methods: In this phase III trial (NCT02856126), patients with advanced HCC were enrolled. Eligible patients were randomly assigned (2:1) to receive Sorafenib (400mg orally twice daily) plus HAIC with oxaliplatin & fluorouracil per 3 weeks or TACE (50mg of epirubicin and lobaplatin mixed with lipiodol and polyvinyl alcohol) until disease progression or unacceptable toxicity. Primary endpoint was overall survival (OS). Whole-Exome Sequencing and RNA Sequencing of tumor biopsy samples were performed for predictive biomarker exploration (cohort 1) and validation (SHATA-001). Results: Between August 4, 2016, and October 25, 2020, 207 patients were assigned to treatment (SoraHAIC, n=141; SoraTACE, n=66). Primary endpoint was met, with the median OS of 15.6 (95% CI, 12.7-18.6) months in SoraHAIC group and 11.2 (7.6-14.6) months in SoraTACE group (HR, 0.53 [0.38-0.74], p<0.001). Median progression-free survival was 6.9 (5.4-8.5) months and 4.2 (2.9-5.3) months in the respective groups (HR, 0.51[0.37-0.69], p<0.001). Safety analyses (n=200) showed frequencies of grade 3-4 elevated alanine aminotransferase (5% vs 21.7%; p<0.001) and elevated aspartate aminotransferase (16.4% vs 35%; p<0.001) significantly higher with SoraTACE than SoraHAIC. Serious adverse events occurred more frequently with SoraTACE (56.7%) than SoraHAIC (39.3%) (p=0.023). In cohort 1, we analyzed the metabolic activity of fatty acid degradation (FAD) pathway using FAD score, which was positively correlated with longer OS, especially in patients received TACE. Furthermore, we found a down regulation of FAD activity in patients enrolled in SHATA-001. And higher FAD score showed a better tumor response to SoraTACE, while FAD had no difference between responders and non-responders in SoraHAIC. Conclusions: This trial demonstrated SoraHAIC significantly improved OS over SoraTACE in patients with advanced HCC. However, patients with high FAD activity might have a better tumor response and survival benefit to SoraTACE than SoraHAIC. Clinical trial information: NCT02856126.