Abstract Background The goal of this study was to understand gene expression signatures of hepatocellular carcinoma (HCC) recurrence in subjects with hepatitis C virus (HCV) infection. Recurrence-free survival (RFS) following curative resection of HCC in subjects with HCV is highly variable. Traditional clinico-pathological endpoints are recognized as weak predictors of RFS. It has been suggested that gene expression profiling of HCC and nontumoral liver tissue may improve prediction of RFS, aid in understanding of the underlying liver disease, and guide individualized patient management. Frozen samples of the tumors and nontumoral liver were obtained from 47 subjects with HCV-associated HCC. Additional nontumoral liver samples were obtained from HCV-free subjects with metastatic liver tumors. Gene expression profiling data was used to determine the molecular signature of HCV-associated HCC and to develop a predictor of RFS. Results The molecular profile of the HCV-associated HCC confirmed central roles for MYC and TGFβ1 in liver tumor development. Gene expression in tumors was found to have poor predictive power with regards to RFS, but analysis of nontumoral tissues yielded a strong predictor for RFS in late-recurring (>1 year) subjects. Importantly, nontumoral tissue-derived gene expression predictor of RFS was highly significant in both univariable and multivariable Cox proportional hazard model analyses. Conclusions Microarray analysis of the nontumoral tissues from subjects with HCV-associated HCC delivers novel molecular signatures of RFS, especially among the late-recurrence subjects. The gene expression predictor may hold important insights into the pathobiology of HCC recurrence and de novo tumor formation in cirrhotic patients.