KEYNOTE-033 (NCT02864394) was a multicountry, open-label, phase 3 study that compared pembrolizumab vs docetaxel in previously treated, programmed death-ligand 1 (PD-L1)-positive, advanced non-small cell lung cancer (NSCLC), with most patients enrolled in mainland China. Eligible patients were randomized (1:1) to pembrolizumab 2 mg/kg or docetaxel 75 mg/m2 every 3 weeks. Primary endpoints were overall survival (OS) and progression-free survival and were evaluated sequentially using stratified log-rank tests, first in patients with PD-L1 tumor proportion score (TPS) ≥50% and then in patients with PD-L1 TPS ≥1% (significance threshold: P < .025, one-sided). A total of 425 patients were randomized to pembrolizumab (N = 213) or docetaxel (N = 212) between 8 September 2016 and 17 October 2018. In patients with a PD-L1 TPS ≥50% (n = 227), median OS was 12.3 months with pembrolizumab and 10.9 months with docetaxel; the hazard ratio (HR) was 0.83 (95% confidence interval [CI]: 0.61-1.14; P = .1276). Because the significance threshold was not met, sequential testing of OS and PFS was ceased. In patients with a PD-L1 TPS ≥1%, the HR for OS for pembrolizumab vs docetaxel was 0.75 (95% CI: 0.60-0.95). In patients from mainland China (n = 311) with a PD-L1 TPS ≥1%, HR for OS was 0.68 (95% CI: 0.51-0.89). Incidence of grade 3 to 5 treatment-related AEs was 11.3% with pembrolizumab vs 47.5% with docetaxel. In summary, pembrolizumab improved OS vs docetaxel in previously treated, PD-L1-positive NSCLC without unexpected safety signals; although the statistical significance threshold was not reached, the numerical improvement is consistent with that previously observed for pembrolizumab in previously treated, advanced NSCLC.