Immunotherapy with checkpoint inhibitors, allowing recovery of effector cells function, has demonstrated to be highly effective in many tumor types and represents a true revolution in oncology. Recently, the anti-PD1 agent pembrolizumab was granted FDA approval for the first line treatment of patients with advanced non-small cell lung cancer (NSCLC) whose tumors show PD-L1 expression in ≥ 50% of neoplastic cells and as a second line treatment for patients with NSCLC expressing PD-L1 in ≥1% of neoplastic cells, evaluated with a validated assay. For the large majority of patients such evaluation is made on small biopsies. However, small tissue samples such as core biopsies might not be representative of tumors and may show divergent results given the possible heterogeneous immunoexpression of the biomarker. We therefore sought to evaluate PD-L1 expression concordance in a cohort of 239 patients using tissue microarrays (TMA) as surrogates of biopsies stained with a validated PD-L1 immunohistochemical assay (SP263) and report the degree of discordance among tissue cores in order to understand how such heterogeneity could affect decisions regarding therapy. We observed a discordance rate of 20% and 7.9% and a Cohen's κ value of 0.53 (moderate) and 0,48 (moderate) for ≥ 1% and ≥ 50% cutoffs, respectively. Our results suggest that caution must be taken when evaluating single biopsies from patients with advanced NSCLC eligible for immunotherapy; moreover, at least 4 biopsies are necessary in order to minimize the risk of tumor misclassification.