Programmed death ligand 1 (PD-L1) protein expression by immunohistochemistry is a promising biomarker for PD-1/PD-L1 blockade in hepatocellular carcinoma. There are a number of commercially available PD-L1 assays. Our study aimed to compare the analytical performance of different PD-L1 assays and evaluate the reliability of pathologists in PD-L1 scoring. Consecutive sections from tumor samples from 55 patients with surgically resected primary hepatocellular carcinoma were stained with four standardized PD-L1 assays (22C3, 28-8, SP142, and SP263). We also correlated the PD-L1 protein level by immunohistochemistry with the mRNA level of those genes associated with tumor immune microenvironment by the NanoString platform. Five pathologists independently assessed PD-L1 expression on tumor cells [tumor proportion score] together with tumor-infiltrating immune cells (combined positive score). The 22C3, 28-8, and SP263 assays had comparable sensitivity in detecting PD-L1 expression, whereas the SP142 assay was the least sensitive assay. The inter-assay agreement measured by intraclass correlation coefficients for the tumor proportion score and combined positive score were 0.646 and 0.780, respectively. The inter-rater agreement was good to excellent (the overall intraclass correlation coefficient for the tumor proportion score and combined positive score was 0.946 and 0.809, respectively). Pathologists were less reliable in scoring combined positive score than tumor proportion score, particularly when using the SP142 assay. Up to 18% of samples were misclassified by individual pathologists in comparison to the consensus score at the cutoff of combined positive score ≥ 1. The combined positive score by the 22C3 assay demonstrated the strongest correlation with immune-related gene mRNA signatures, closely followed by combined positive scores by the 28-8 and SP263 assays. In conclusion, the 22C3, 28-8, and SP263 assays are highly concordant in PD-L1 scoring and suggest the interchangeability of these three assays. Further improvement of the accuracy in assessing PD-L1 expression at a low cutoff is still necessary.