Background and Objective Immune checkpoint inhibition has shed light on a new era in cancer therapy, and randomized clinical trials have demonstrated that a meaningful portion of the overall population of metastatic gastric cancer (GC) patients may derive clinical benefit from immunotherapy, which raises the relevance in identifying predictive biomarkers. Programmed cell death-ligand 1 (PD-L1) expression has demonstrated a significant association between level of expression and the magnitude of benefit derived from immune checkpoint inhibition in GC. Nevertheless, this biomarker shows several pitfalls that must be considered in the therapeutic decision to incorporate immune checkpoint inhibition as the standard of care of GC, such as spatial and temporal heterogeneity, interobserver variability, immunohistochemistry (IHC) assay, and influence by chemotherapy or radiation therapy. Methods In the present comprehensive review, we revised the main studies regarding PD-L1 evaluation in GC. Key Content and Findings Here we describe the molecular characteristics of the tumor microenvironment in GC, the obstacles in the interpretation of PD-L1 expression and present the data of the clinical trials that have evaluated the efficacy and safety of immune checkpoint inhibition and the association with the biomarker expression, both in first-line and later lines of therapy. Conclusions From the emerging predictive biomarkers for immune checkpoint inhibition, PD-L1 has demonstrated a meaningful association between level of expression in tumor microenvironment and the magnitude of benefit derived from immune checkpoint inhibition in GC.