The optimal perioperative chemotherapy for patients with muscle-invasive bladder cancer is not defined. The VESPER (French Genito-Urinary Tumor Group and French Association of Urology V05) trial reported improved 3-year progression-free survival with dose-dense methotrexate, vinblastine, doxorubicin and cisplatin (dd-MVAC) versus gemcitabine and cisplatin (GC) in patients who received neoadjuvant therapy, but not in the overall perioperative setting. In this Article, we report on the secondary endpoints of overall survival and time to death due to bladder cancer at 5-year follow-up. VESPER was an open-label, randomised, phase 3 trial done at 28 university hospitals or comprehensive cancer centres in France, in which adults (age ≤18 years and ≤80 years) with primary bladder cancer and histologically confirmed muscle-invasive urothelial carcinoma were randomly allocated (1:1; block size four) to treatment with dd-MVAC (every 2 weeks for a total of six cycles) or GC (every 3 weeks for a total of four cycles). Overall survival and time to death due to bladder cancer (presented as 5-year cumulative incidence of death due to bladder cancer) was analysed by intention to treat (ITT) in all randomly assigned patients. Overall survival was assessed by the Kaplan-Meier method with the treatment groups compared with log-rank test stratified for mode of administration of chemotherapy (neoadjuvant or adjuvant) and lymph node involvement. Time to death due to bladder cancer was analysed with an Aalen model for competing risks and a Fine and Gray regression model stratified for the same two covariates. Results were presented for the total perioperative population and for the neoadjuvant and adjuvant subgroups. The trial is registered with ClinicalTrials.gov, NCT01812369, and is complete. From Feb 25, 2013, to March 1, 2018, 500 patients were randomly assigned, of whom 493 were included in the final ITT population (245 [50%] in the GC group and 248 [50%] in the dd-MVAC group; 408 [83%] male and 85 [17%] female). 437 (89%) patients received neoadjuvant chemotherapy. Median follow-up was 5·3 years (IQR 5·1-5·4); 190 deaths at the 5-year cutoff were reported. In the perioperative setting (total ITT population), we found no evidence of association of overall survival at 5 years with dd-MVAC treatment versus GC treatment (64% [95% CI 58-70] vs 56% [50-63], stratified hazard ratio [HRstrat] 0·79 [95% CI 0·59-1·05]). Time to death due to bladder cancer was increased in the dd-MVAC group compared with in the GC group (5-year cumulative incidence of death: 27% [95% CI 21-32] vs 40% [34-46], HRstrat 0·61 [95% CI 0·45-0·84]). In the neoadjuvant subgroup, overall survival at 5 years was improved in the dd-MVAC group versus the GC group (66% [95% CI 60-73] vs 57% [50-64], HR 0·71 [95% CI 0·52-0·97]), as was time to death due to bladder cancer (5-year cumulative incidence: 24% [18-30] vs 38% [32-45], HR 0·55 [0·39-0·78]). In the adjuvant subgroup, the results were not conclusive due to the small sample size. Bladder cancer progression was the cause of death for 157 (83%) of the 190 deaths; other causes of death included cardiovascular events (eight [4%] deaths), deaths related to chemotherapy toxicity (four [2%]), and secondary cancers (four [2%]). Our results on overall survival at 5 years were in accordance with the primary endpoint analysis (3-year progression-free survival). We found no evidence of improved overall survival with dd-MVAC over GC in the perioperative setting, but the data support the use of six cycles of dd-MVAC over four cycles of GC in the neoadjuvant setting. These results should impact practice and future trials of immunotherapy in bladder cancer. French National Cancer Institute.