Immune checkpoint inhibitors (ICI) have been applied to treat advanced stage hepatocellular carcinoma (HCC) and obtain promising effects. However, tumor response to treatment was unpredictable. A predicting biomarker of objective response or disease-control is an unmet need for patient selection. In this study, 45 advanced HCC patients who failed to sorafenib treatment and received nivolumab, 3 mg/kg bi-weekly, were included. Tumor responses to nivolumab treatment were assessed by the modified response evaluation criteria in solid tumors (mRECIST) criteria. Tumor responses were correlated to clinical characteristics to find out response predictors. In this small series, the prevalence of extrahepatic nodal metastasis, distant metastasis, and portal vein thrombus among the patients were 22.2% (n = 10), 48.9% (n = 22), and 42.2% (n = 19), respectively. The pre-treatment tumor size was 7.2 ± 4.2 cm in maximal diameter, and the calculated total tumor volume was 619.0 ± 831.1 cm3. Among 45 patients, 3 patients had partial response (PR), 11 had stable disease (SD), and the other 31 had progression of disease. By correlating clinical data to the patients with PR and SD, serum neutrophil-to-lymphocyte ratio (NLR) (hazard ratio (HR) = 2.04) and patient-generated subjective global assessment (PG-SGA) score (HR = 2.30) were the independent factors in multivariate analysis. By receiver operating characteristic curve analysis, pre-treatment NLR ≤ 2.5 and PG-SGA score < 4 were the cutoff points to predict tumor response to ICI treatment. In conclusion, biomarkers to predict tumor response for HCC are still lacking in this costly ICI therapy. In this study, NLR ≤ 2.5 and PG-SGA score < 4 indicated disease-control, and can be applied as biomarkers to select the right patients to receive this costly therapy.