BACKGROUND AND AIMS: Molecular classification is a promising tool for prognosis prediction and optimizing precision therapy for hepatocellular carcinoma (HCC). Here, we aimed to develop a molecular classification of HCC based on the fatty acid degradation (FAD) pathway, fully characterize it, and evaluate its ability in guiding personalized therapy. APPROACH AND RESULTS: We performed RNA sequencing (RNA-seq), PCR-array, lipidomics, metabolomics, and proteomics analysis of 41 HCC patients, in which 17 patients received anti-PD-1 therapy. Single-cell RNA sequencing (scRNA-seq) was performed to explore the tumor microenvironment. Nearly 60 publicly-available multi-omics datasets were analyzed. The associations between FAD subtypes and response to sorafenib, transarterial chemoembolization (TACE), immune checkpoint inhibitor (ICI) were assessed in patient cohorts, patient-derived xenograft (PDX), and spontaneous mice models. A novel molecular classification named F subtype (F1, F2, and F3) was identified based on the FAD pathway, distinguished by clinical, mutational, epigenetic, metabolic, and immunological characteristics. F1 subtypes exhibited high infiltration with immunosuppressive microenvironment. Subtype-specific therapeutic strategies were identified, in which F1 subtypes with the lowest FAD activities represent responders to compounds YM-155 and Alisertib, sorafenib, anti-PD1, anti-PD-L1, and atezolizumab plus bevacizumab (T + A) treatment, while F3 subtypes with the highest FAD activities are responders to TACE. F2 subtypes, the intermediate status between F1 and F3, are potential responders to T + A combinations. We provide preliminary evidence that the FAD subtypes can be diagnosed based on liquid biopsies. CONCLUSIONS: We identified three FAD subtypes with unique clinical and biological characteristics, which could optimize individual cancer patient therapy and help clinical decision-making.