Neoadjuvant or adjuvant immunotherapy can improve outcomes in patients with resectable non-small-cell lung cancer (NSCLC). Perioperative regimens may combine benefits of both to improve long-term outcomes.We randomly assigned patients with resectable NSCLC (stage II to IIIB [N2 node stage] according to the eighth edition of the AJCC Cancer Staging Manual) to receive platinum-based chemotherapy plus durvalumab or placebo administered intravenously every 3 weeks for 4 cycles before surgery, followed by adjuvant durvalumab or placebo intravenously every 4 weeks for 12 cycles. Randomization was stratified according to disease stage (II or III) and programmed death ligand 1 (PD-L1) expression (≥1% or <1%). Primary end points were event-free survival (defined as the time to the earliest occurrence of progressive disease that precluded surgery or prevented completion of surgery, disease recurrence [assessed in a blinded fashion by independent central review], or death from any cause) and pathological complete response (evaluated centrally).A total of 802 patients were randomly assigned to receive durvalumab (400 patients) or placebo (402 patients). The duration of event-free survival was significantly longer with durvalumab than with placebo; the stratified hazard ratio for disease progression, recurrence, or death was 0.68 (95% confidence interval [CI], 0.53 to 0.88; P = 0.004) at the first interim analysis. At the 12-month landmark analysis, event-free survival was observed in 73.4% of the patients who received durvalumab (95% CI, 67.9 to 78.1), as compared with 64.5% of the patients who received placebo (95% CI, 58.8 to 69.6). The incidence of pathological complete response was significantly greater with durvalumab than with placebo (17.2% vs. 4.3% at the final analysis; difference, 13.0 percentage points; 95% CI, 8.7 to 17.6; P<0.001 at interim analysis of data from 402 patients). Event-free survival and pathological complete response benefit were observed regardless of stage and PD-L1 expression. Adverse events of maximum grade 3 or 4 occurred in 42.4% of patients with durvalumab and in 43.2% with placebo. Data from 62 patients with documented EGFR or ALK alterations were excluded from the efficacy analyses in the modified intention-to-treat population.In patients with resectable NSCLC, perioperative durvalumab plus neoadjuvant chemotherapy was associated with significantly greater event-free survival and pathological complete response than neoadjuvant chemotherapy alone, with a safety profile that was consistent with the individual agents. (Funded by AstraZeneca; AEGEAN ClinicalTrials.gov number, NCT03800134.).