Checkpoint inhibitors directed against programmed death receptor 1 (PD-1) and its ligand (PD-L1) changed the treatment of advanced lung non–small cell carcinomas. The decision to treat patients is influenced by PD-L1 expression by tumor cells, but evidence indicates that this staining is heterogenous within a tumor. As PD-L1 staining is tested mostly on biopsies, false negative results can occur due to sampling issues. The clinical impact of this heterogeneity has not been established. We selected 241 patients who underwent pulmonary resection for adenocarcinoma. Tissue microarrays were constructed with five 1 mm cores representative of the histologic patterns observed in each tumor and stained for PD-L1. For each core, the histologic pattern and the percentage of PD-L1 positive tumor cells were noted. Staining heterogeneity was defined as cases with both positive and negative cores at positivity thresholds of 1%, 10%, and 50% of tumor cells. At the 50% cut-off, 37.8% of patients were PD-L1 positive, whereas 22.4% showed staining heterogeneity. Among patients with 1 negative core, 26.5% also had a positive core and could have been misclassified based on 1 biopsy. Mean staining of PD-L1 was higher in solid (47.9%) and micropapillary (24.2%) patterns and was lower in acinar (14.1%), papillary (3.4%), and lepidic (6.4%) architectures. A significant proportion of patients presented a heterogenous staining for PD-L1. A total of 26.5% of patients negative on 1 core turned out to be positive on another core, which raises the consideration of rebiopsy, in particular when lepidic, acinar, or papillary patterns are observed on a biopsy.