Background: Atezolizumab can induce sustained responses in metastatic urothelial carcinoma. We report the results of IMvigor130, a phase 3 trial that compared atezolizumab with or without platinum-based chemotherapy versus placebo plus platinum-based chemotherapy in first-line metastatic urothelial carcinoma. Methods: In this multicentre, phase 3, randomised trial, untreated patients aged 18 years or older with locally advanced or metastatic urothelial carcinoma, from 221 sites in 35 countries, were randomly assigned to receive atezolizumab plus platinum-based chemotherapy (group A), atezolizumab monotherapy (group B), or placebo plus platinum-based chemotherapy (group C). Patients received 21-day cycles of gemcitabine (1000 mg/m2 body surface area, administered intravenously on days 1 and 8 of each cycle), plus either carboplatin (area under the curve of 4·5 mg/mL per min administered intravenously) or cisplatin (70 mg/m2 body surface area administered intravenously) on day 1 of each cycle with either atezolizumab (1200 mg administered intravenously on day 1 of each cycle) or placebo. Group B patients received 1200 mg atezolizumab, administered intravenously on day 1 of each 21-day cycle. The co-primary efficacy endpoints for the intention-to-treat population were investigator-assessed Response Evaluation Criteria in Solid Tumours 1.1 progression-free survival and overall survival (group A vs group C) and overall survival (group B vs group C), which was to be formally tested only if overall survival was positive for group A versus group C. The trial is registered with ClinicalTrials.gov, NCT02807636. Findings: Between July 15, 2016, and July 20, 2018, we enrolled 1213 patients. 451 (37%) were randomly assigned to group A, 362 (30%) to group B, and 400 (33%) to group C. Median follow-up for survival was 11·8 months (IQR 6·1–17·2) for all patients. At the time of final progression-free survival analysis and interim overall survival analysis (May 31, 2019), median progression-free survival in the intention-to-treat population was 8·2 months (95% CI 6·5–8·3) in group A and 6·3 months (6·2–7·0) in group C (stratified hazard ratio [HR] 0·82, 95% CI 0·70–0·96; one-sided p=0·007). Median overall survival was 16·0 months (13·9–18·9) in group A and 13·4 months (12·0–15·2) in group C (0·83, 0·69–1·00; one-sided p=0·027). Median overall survival was 15·7 months (13·1–17·8) for group B and 13·1 months (11·7–15·1) for group C (1·02, 0·83–1·24). Adverse events that led to withdrawal of any agent occurred in 156 (34%) patients in group A, 22 (6%) patients in group B, and 132 (34%) patients in group C. 50 (11%) patients in group A, 21 (6%) patients in group B, and 27 (7%) patients in group C had adverse events that led to discontinuation of atezolizumab or placebo. Interpretation: Addition of atezolizumab to platinum-based chemotherapy as first-line treatment prolonged progression-free survival in patients with metastatic urothelial carcinoma. The safety profile of the combination was consistent with that observed with the individual agents. These results support the use of atezolizumab plus platinum-based chemotherapy as a potential first-line treatment option for metastatic urothelial carcinoma. Funding: F Hoffmann-La Roche and Genentech.