BACKGROUND: Despite multimodal therapy, 5-year overall survival for locally advanced head and neck squamous cell carcinoma (HNSCC) is about 50%. We assessed the addition of pembrolizumab to concurrent chemoradiotherapy for locally advanced HNSCC. METHODS: In the randomised, double-blind, phase 3 KEYNOTE-412 trial, participants with newly diagnosed, high-risk, unresected locally advanced HNSCC from 130 medical centres globally were randomly assigned (1:1) to pembrolizumab (200 mg) plus chemoradiotherapy or placebo plus chemoradiotherapy. Randomisation was done using an interactive response technology system and was stratified by investigator's choice of radiotherapy regimen, tumour site and p16 status, and disease stage, with participants randomly assigned in blocks of four per stratum. Participants, investigators, and sponsor personnel were masked to treatment assignments. Local pharmacists were aware of assignments to support treatment preparation. Pembrolizumab and placebo were administered intravenously once every 3 weeks for up to 17 doses (one before chemoradiotherapy, two during chemoradiotherapy, 14 as maintenance therapy). Chemoradiotherapy included cisplatin (100 mg/m) administered intravenously once every 3 weeks for two or three doses and accelerated or standard fractionation radiotherapy (70 Gy delivered in 35 fractions). The primary endpoint was event-free survival analysed in all randomly assigned participants. Safety was analysed in all participants who received at least one dose of study treatment. This study is registered with ClinicalTrials.gov, NCT03040999, and is active but not recruiting. FINDINGS: Between April 19, 2017, and May 2, 2019, 804 participants were randomly assigned to the pembrolizumab group (n=402) or the placebo group (n=402). 660 (82%) of 804 participants were male, 144 (18%) were female, and 622 (77%) were White. Median study follow-up was 47·7 months (IQR 42·1-52·3). Median event-free survival was not reached (95% CI 44·7 months-not reached) in the pembrolizumab group and 46·6 months (27·5-not reached) in the placebo group (hazard ratio 0·83 [95% CI 0·68-1·03]; log-rank p=0·043 [significance threshold, p≤0·024]). 367 (92%) of 398 participants treated in the pembrolizumab group and 352 (88%) of 398 participants treated in the placebo group had grade 3 or worse adverse events. The most common grade 3 or worse adverse events were decreased neutrophil count (108 [27%] of 398 participants in the pembrolizumab group vs 100 [25%] of 398 participants in the placebo group), stomatitis (80 [20%] vs 69 [17%]), anaemia (80 [20%] vs 61 [15%]), dysphagia (76 [19%] vs 62 [16%]), and decreased lymphocyte count (76 [19%] vs 81 [20%]). Serious adverse events occurred in 245 (62%) participants in the pembrolizumab group versus 197 (49%) participants in the placebo group, most commonly pneumonia (43 [11%] vs 25 [6%]), acute kidney injury (33 [8%] vs 30 [8%]), and febrile neutropenia (24 [6%] vs seven [2%]). Treatment-related adverse events led to death in four (1%) participants in the pembrolizumab group (one participant each from aspiration pneumonia, end-stage renal disease, pneumonia, and sclerosing cholangitis) and six (2%) participants in the placebo group (three participants from pharyngeal haemorrhage and one participant each from mouth haemorrhage, post-procedural haemorrhage, and sepsis). INTERPRETATION: Pembrolizumab plus chemoradiotherapy did not significantly improve event-free survival compared with chemoradiotherapy alone in a molecularly unselected, locally advanced HNSCC population. No new safety signals were seen. Locally advanced HNSCC remains a challenging disease that requires better treatment approaches. FUNDING: Merck Sharp & Dohme, a subsidiary of Merck & Co, Rahway, NJ, USA.