Background: Pembrolizumab prolongs progression-free and overall survival among patients with advanced melanoma and recurrence-free survival in resected stage III disease. KEYNOTE-716 assessed pembrolizumab as adjuvant therapy in patients with completely resected, high-risk, stage II melanoma. We report results from the planned first and second interim analyses for recurrence-free survival. Methods: In this double-blind, randomised, placebo-controlled phase 3 study, involving 160 academic medical centres and hospitals in 16 countries (Australia, Belgium, Brazil, Canada, Chile, France, Germany, Israel, Italy, Japan, Poland, South Africa, Spain, Switzerland, the UK, and the USA), patients aged 12 years or older with newly diagnosed, completely resected stage IIB or IIC melanoma (TNM stage T3b or T4 with a negative sentinel lymph node biopsy) were recruited. Eligible patients were randomly assigned (1:1), in blocks of four and stratified by T-category (3b, 4a, and 4b) and paediatric status (age 12–17 years vs ≥18 years), using an interactive response technology system to intravenous pembrolizumab 200 mg (2 mg/kg in paediatric patients) or placebo every 3 weeks for 17 cycles or until disease recurrence or unacceptable toxicity. All patients, clinical investigators, and analysts were masked to treatment assignment. The primary endpoint was investigator-assessed recurrence-free survival (defined as time from randomisation to recurrence or death) in the intention-to-treat (ITT) population (ie, all patients randomly assigned to treatment). The primary endpoint was met if recurrence-free survival was significantly improved for pembrolizumab versus placebo at either the first interim analysis (after approximately 128 patients had events) or second interim analysis (after 179 patients had events) under multiplicity control. Safety was assessed in all patients randomly assigned to treatment who received at least one dose of study treatment. This study is registered with ClinicalTrials.gov, NCT03553836, and is closed to accrual. Findings: Between Sept 23, 2018, and Nov 4, 2020, 1182 patients were screened, of whom 976 were randomly assigned to pembrolizumab (n=487) or placebo (n=489; ITT population). The median age was 61 years (IQR 52–69) and 387 (40%) patients were female and 589 (60%) were male. 874 (90%) of 976 patients were White and 799 (82%) were not Hispanic or Latino. 483 (99%) of 487 patients in the pembrolizumab group and 486 (99%) of 489 in the placebo group received assigned treatment. At the first interim analysis (data cutoff on Dec 4, 2020; median follow-up of 14·4 months [IQR 10·2–18·7] in the pembrolizumab group and 14·3 months [10·1–18·7] in the placebo group), 54 (11%) of 487 patients in the pembrolizumab group and 82 (17%) of 489 in the placebo group had a first recurrence of disease or died (hazard ratio [HR] 0·65 [95% CI 0·46–0·92]; p=0·0066). At the second interim analysis (data cutoff on June 21, 2021; median follow-up of 20·9 months [16·7–25·3] in the pembrolizumab group and 20·9 months [16·6–25·3] in the placebo group), 72 (15%) patients in the pembrolizumab group and 115 (24%) in the placebo group had a first recurrence or died (HR 0·61 [95% CI 0·45–0·82]). Median recurrence-free survival was not reached in either group at either assessment timepoint. At the first interim analysis, grade 3–4 treatment-related adverse events occurred in 78 (16%) of 483 patients in the pembrolizumab groups versus 21 (4%) of 486 in the placebo group. At the first interim analysis, four patients died from an adverse event, all in the placebo group (one each due to pneumonia, COVID-19-related pneumonia, suicide, and recurrent cancer), and at the second interim analysis, one additional patient, who was in the pembrolizumab group, died from an adverse event (COVID-19-related pneumonia). No deaths due to study treatment occurred. Interpretation: Pembrolizumab as adjuvant therapy for up to approximately 1 year for stage IIB or IIC melanoma resulted in a significant reduction in the risk of disease recurrence or death versus placebo, with a manageable safety profile. Funding: Merck Sharp & Dohme, a subsidiary of Merck & Co, Kenilworth, NJ, USA.