Background: The COVID-19 pandemic led to Cancer Care Ontario’s approval of extended dosing (ED) of pembrolizumab in the management of metastatic non-small cell lung cancer (mNSCLC) as a means of reducing the risk of exposure through less visits to the cancer center. Pembrolizumab ED has been evaluated through pharmacokinetic studies and simulated model-based analyses, however clinical data are lacking. The aim of this study is to evaluate real-world outcomes and safety of ED vs standard dosing (SD) of pembrolizumab. Methods: This retrospective cohort study included mNSCLC patients (all histologies) from the Cancer Centre of Southeastern Ontario (Kingston, ON, Canada) who received at least one cycle of single agent pembrolizumab with SD (2 mg/kg IV up to a maximum of 200 mg q3 weeks) or ED (4 mg/kg IV up to a maximum of 400 mg q6 weeks), from January 2018 to December 2020. Data were collected from review of clinical, pathological, and radiological reports. Study outcomes included proportion of patients alive at data cutoff, median number of treatment cycles, immune-related adverse events (irAEs), and time to toxicity (TTT). Statistical analyses included Chi-square, Mann-Whitney and t-tests. Results: Of 90 patients evaluated, 18 (20%) switched to ED. The proportion of those alive at data cutoff was significantly higher in the ED vs. SD group (94% vs. 26%; p<0.01). The ED group also received more cycles of therapy (median=19, with 11.5 cycles received before the switch) as compared to the SD group (median=3.5). The median TTT was 130 (ED) vs. 105 (SD) days, and the rate (44% vs. 32%; p=0.407) and severity of irAEs (≥ grade 3 events: 50% vs. 52%) were similar. A higher proportion of ED patients discontinued treatment due to toxicity (28% vs. 14%; p<0.01), and 5/8 (62.5%) of the irAEs from the ED group occurred after switching (all events resulted in treatment discontinuation). Conclusion: A greater proportion of ED patients were alive at the time of data cutoff, but this may reflect a selection bias as they were more likely switched considering disease stability (i.e., more cycles were received prior to switching). The rate and severity of irAEs were similar between the two groups, however a high proportion of irAEs in the ED group occurred following the switch, which justified treatment cessation. These results are limited by small patient numbers, and a possible impact of ED on toxicity should be further validated in larger cohorts.