BACKGROUND & nbsp;Pembro is a standard of care for advanced NSCLC. PEARLS/ KEYNOTE-091 (NCT02504372) evaluated adjuvant pembro vs placebo for patients (pts) with completely resected early-stage NSCLC.& nbsp;METHODS & nbsp;Eligible pts with completely resected stage IB (T > 4 cm), II, or IIIA NSCLC per AJCC v7 followed by adjuvant chemo as indicated per guidelines, ECOG PS 0-1, and any PD-L1 expression were randomized 1:1 to pembro 200 mg or placebo Q3W for 18 doses (w1 year). Dual primary end points are DFS in the all-comers and PD-L1 TPS > 50% populations. Secondary end points are DFS in the TPS > 1% population, OS in the all-comers, TPS > 50% and > 1% populations, and safety. Data are from the protocol-specified second interim analysis (IA2; 20 September 2021, data cutoff).& nbsp;RESULTS & nbsp;1177 pts were randomized to pembro (N = 590) or placebo (N = 587), including 168 and 165, respectively, who had TPS >= 50%. Baseline characteristics were generally balanced between arms. Median time from randomization to data cutoff for IA2 was 35.6 mo (range 16.5-68.0). DFS was significantly improved with pembro in the all-comers population (median 53.6 vs 42.0 mo; HR 0.76; 95% CI 0.63-0.91; P = 0.0014); the significance boundary was not crossed for the TPS >= 50% population (median not reached in either arm; HR 0.82; 95% CI 0.57-1.18; P = 0.14). With only 209 events, the significance boundary for OS in the all-comers population was not crossed (18-mo rate 91.7% vs 91.3%; HR 0.87; 95% CI 0.67-1.15; P = 0.17). Median number of doses was 17 for pembro vs 18 for placebo. AEs were grade >= 3 in 34.1% of pts in the pembro arm vs 25.8% in the placebo arm and led to discontinuation in 19.8% vs 5.9%; treatment-related AEs led to death in 0.7% vs 0%.& nbsp;CONCLUSIONS & nbsp;Adjuvant pembro provided a statistically significant, clini-cally meaningful DFS improvement following complete resection and, when indicated, adjuvant chemo in pts with stage IB (T >= 4 cm)-IIIA NSCLC, regardless of PD-L1 expres-sion. The pembro safety profile was as expected. DFS in the PD-L1-defined populations and OS will be tested at future analyses according to the analysis plan.