BACKGROUND: Evidence for the efficacy of combined PD-1 and HER2 blockade with chemotherapy on progression-free and overall survival in HER2-positive gastro-oesophageal cancer is scarce. The first interim analysis of the randomised, phase 3 KEYNOTE-811 study showed a superior objective response with pembrolizumab compared with placebo when added to trastuzumab plus fluoropyrimidine and platinum-based chemotherapy. Here, we report results from protocol-specified subsequent interim analyses of KEYNOTE-811. METHODS: The randomised, phase 3 KEYNOTE-811 trial involved 168 medical centres in 20 countries worldwide. Patients aged 18 years or older with locally advanced or metastatic HER2-positive gastro-oesophageal junction adenocarcinoma, without previous first-line treatment, were randomly assigned (1:1) by an integrated interactive voice-response and web-response system to intravenous pembrolizumab 200 mg or placebo, both to be combined with standard chemotherapy (fluoropyrimidine and platinum-based therapy) plus trastuzumab every 3 weeks for up to 35 cycles or until disease progression, unacceptable toxic effects, or investigator or participant-initiated withdrawal. Randomisation used a block size of four and was stratified by region, PD-L1 status, and chemotherapy. Dual primary endpoints were progression-free and overall survival, analysed by intention to treat. Safety was assessed in all randomly assigned patients who received at least one dose of study treatment according to the treatment received. KEYNOTE-811 is registered with ClinicalTrials.gov (NCT03615326) and is active but not recruiting. FINDINGS: Between Oct 5, 2018, and Aug 6, 2021, 698 patients were assigned to pembrolizumab (n=350) or placebo (n=348). 564 (81%) were male and 134 (19%) were female. At the third interim analysis, 286 (82%) of 350 patients in the pembrolizumab group and 304 (88%) of 346 in the placebo group who received treatment had discontinued treatment, mostly due to disease progression. At the second interim analysis (median follow-up 28·3 months [IQR 19·4-34·3] in the pembrolizumab group and 28·5 months [20·1-34·3] in the placebo group), median progression-free survival was 10·0 months (95% CI 8·6-11·7) in the pembrolizumab group versus 8·1 months (7·0-8·5) in the placebo group (hazard ratio [HR] 0·72, 95% CI 0·60-0·87; p=0·0002). Median overall survival was 20·0 months (17·8-23·2) versus 16·9 months (15·0-19·8; HR 0·87 [0·72-1·06]; p=0·084). At the third interim analysis (median follow-up 38·4 months [IQR 29·5-44·4] in the pembrolizumab group and 38·6 months [30·2-44·4] in the placebo group), median progression-free survival was 10·0 months (8·6-12·2) versus 8·1 months (7·1-8·6; HR 0·73 [0·61-0·87]), and median overall survival was 20·0 months (17·8-22·1) versus 16·8 months (15·0-18·7; HR 0·84 [0·70-1·01]), but did not meet prespecified criteria for significance and will continue to final analysis. Grade 3 or worse treatment-related adverse events occurred in 204 (58%) of 350 patients in the pembrolizumab group versus 176 (51%) of 346 patients in the placebo group. Treatment-related adverse events that led to death occurred in four (1%) patients in the pembrolizumab group and three (1%) in the placebo group. The most common treatment-related adverse events of any grade were diarrhoea (165 [47%] in the pembrolizumab group vs 145 [42%] in the placebo group), nausea (154 [44%] vs 152 [44%]), and anaemia (109 [31%] vs 113 [33%]). INTERPRETATION: Compared with placebo, pembrolizumab significantly improved progression-free survival when combined with first-line trastuzumab and chemotherapy for metastatic HER2-positive gastro-oesophageal cancer, specifically in patients with tumours with a PD-L1 combined positive score of 1 or more. Overall survival follow-up is ongoing and will be reported at the final analysis. FUNDING: Merck Sharp & Dohme.