9026 Background: The anti–PD-1 antibody pembro (MK-3475) is approved in the US for treating PD-L1–positive NSCLC that progressed after platinum-containing chemotherapy. This approval was based on data from the large, phase 1b KEYNOTE-001 study (NCT01295827). We present updated, long-term OS data for treatment-naive and previously treated patients (pts) enrolled in KEYNOTE-001. Methods: 550 pts received pembro 2 or 10 mg/kg Q3W or 10 mg/kg Q2W until intolerable toxicity, progression, or investigator decision. PD-L1 was assessed by IHC using the 22C3 antibody, with positivity defined as PD-L1 expression on ≥1% of tumor cells (tumor proportion score [TPS] ≥1%). Response was assessed by RECIST v1.1 every 9 wk. Survival was assessed every 2 mo after discontinuation. Results: As of Sep 18, 2015, median follow-up duration was 23.1 mo. Median OS (95% CI) was 22.1 mo (17.1-27.2) for treatment-naive pts and 10.6 mo (8.6-13.3) for previously treated pts. 18-mo OS rates were 58.2% and 37.0% respectively; 24-mo rates were 44.5% and 31.3%. OS increased with increasing PD-L1 TPS (Table). OS by smoking history, histology, and EGFR status is shown (Table). Conclusions: Pembrolizumab provides long-term OS benefit for PD-L1–positive treatment-naive and previously treated NSCLC. Along with data from KEYNOTE-010, these data support both PD-L1 as a predictive biomarker for pembro and the benefit of pembro in pts with PD-L1–positive (TPS ≥1%) NSCLC. The 22.1-mo median OS in treatment-naive pts with PD-L1–positive tumors is very promising and compares favorably with that of standard-of-care chemotherapy. Clinical trial information: NCT01295827.