BACKGROUND: Cisplatin-based adjuvant chemotherapy is the standard of care for patients with resected stage II-IIIA non-small-cell lung cancer (NSCLC). RADIANT and SELECT trial data suggest patients with EGFR-mutant stage IB-IIIA resected NSCLC could benefit from adjuvant EGFR tyrosine kinase inhibitor treatment. We aimed to compare the efficacy of adjuvant gefitinib versus vinorelbine plus cisplatin in patients with completely resected EGFR-mutant stage II-IIIA (N1-N2) NSCLC. METHODS: We did a randomised, open-label, phase 3 trial at 27 centres in China. We enrolled patients aged 18-75 years with completely resected (R0), stage II-IIIA (N1-N2), EGFR-mutant (exon 19 deletion or exon 21 Leu858Arg) NSCLC. Patients were stratified by N stage and EGFR mutation status and randomised (1:1) by Pocock and Simon minimisation with a random element to either gefitinib (250 mg once daily) for 24 months or intravenous vinorelbine (25 mg/m on days 1 and 8) plus intravenous cisplatin (75 mg/m on day 1) every 3 weeks for four cycles. The primary endpoint was disease-free survival in the intention-to-treat population, which comprised all randomised patients; the safety population included all randomised patients who received at least one dose of study medication. Enrolment to the study is closed but survival follow-up is ongoing. The study is registered with ClinicalTrials.gov, number NCT01405079. FINDINGS: Between Sept 19, 2011, and April 24, 2014, 483 patients were screened and 222 patients were randomised, 111 to gefitinib and 111 to vinorelbine plus cisplatin. Median follow-up was 36·5 months (IQR 23·8-44·8). Median disease-free survival was significantly longer with gefitinib (28·7 months [95% CI 24·9-32·5]) than with vinorelbine plus cisplatin (18·0 months [13·6-22·3]; hazard ratio [HR] 0·60, 95% CI 0·42-0·87; p=0·0054). In the safety population, the most commonly reported grade 3 or worse adverse events in the gefitinib group (n=106) were raised alanine aminotransferase and asparate aminotransferase (two [2%] patients with each event vs none with vinorelbine plus cisplatin). In the vinorelbine plus cisplatin group (n=87), the most frequently reported grade 3 or worse adverse events were neutropenia (30 [34%] patients vs none with gefitinib), leucopenia (14 [16%] vs none), and vomiting (eight [9%] vs none). Serious adverse events were reported for seven (7%) patients who received gefitinib and 20 (23%) patients who received vinorelbine plus cisplatin. No interstitial lung disease was noted with gefitinib. No deaths were treatment related. INTERPRETATION: Adjuvant gefitinib led to significantly longer disease-free survival compared with that for vinorelbine plus cisplatin in patients with completely resected stage II-IIIA (N1-N2) EGFR-mutant NSCLC. Based on the superior disease-free survival, reduced toxicity, and improved quality of life, adjuvant gefitinib could be a potential treatment option compared with adjuvant chemotherapy in these patients. However, the duration of benefit with gefitinib after 24 months might be limited and overall survival data are not yet mature. FUNDING: Guangdong Provincial Key Laboratory of Lung Cancer Translational Medicine; National Health and Family Planning Commission of People's Republic of China; Guangzhou Science and Technology Bureau; AstraZeneca China.