Introduction:The benefits of combining transarterial chemoembolization (TACE) and sorafenib (TACE-S) over TACE alone for treatment of unresectable hepatocellular carcinoma (HCC) remain controversial. Yet, such populations are heterogeneous in terms of baseline characteristics.Objective:To investigate the predictors of survival benefits from added sorafenib and identify the potential candidates for TACE-S.Methods:This multicenter observational study was conducted in 17 Chinese tertiary hospitals for patients with unresectable, liver-confined HCC. Eligible patients with performance status score of <= 1 and Child-Pugh score of <= 7 were treated with TACE or TACE-S. Interactions between treatment and baseline variables were evaluated to find indicators for survival benefits, based on which the patients were stratified. Multivariate models adjusted for baseline characteristics or propensity score were used to compare overall survival (OS) and time to tumor progression (TTP).Results:From January 2009 to December 2015, 1,719 consecutive patients received TACE (n= 1,406) or TACE-S (n= 313). Although TACE-S compared with TACE improved TTP (adjusted hazard ratio [HR] 0.75,p= 0.008), no difference in OS was observed (adjusted HR 0.87,p= 0.090). Nevertheless, the tumor burden (sum of maximum diameter of largest tumor [cm] and tumor number) and albumin-bilirubin (ALBI) score independently predicted the survival benefits from added sorafenib (interactionp< 0.001). For patients with either moderate tumor burden (7-13) or low ALBI score (no more than -2.8) defined as candidates, TACE-S prolonged OS (adjusted HR 0.73,p= 0.003) and TTP (adjusted HR 0.72,p= 0.014) compared to TACE alone, whereas its superiority disappeared in non-candidates.Conclusions:Not all unresectable HCC patients but those with moderate tumor burden or low ALBI score achieve survival benefits from TACE-S compared with TACE alone. Future randomized controlled trials focusing on the subset are warranted.