Background: The present study (VINGEM) is the first randomised trial comparing vinflunine/gemcitabine (VG) to standard carboplatin/gemcitabine (CG) in patients with advanced urothelial carcinoma (aUC) ineligible for treatment with cisplatin. Patients and methods: Patients with aUC, creatinine clearance 30–60 ml/min, performance status ≤1 and no prior chemotherapy for metastatic disease were randomised to the experimental arm (vinflunine 280 or 250 mg/m2 day 1, gemcitabine 1000 mg/m2 days 1 and 8, q21 days) or the control arm (carboplatin AUC 4.5 day 1, gemcitabine 1000 mg/m2 days 1 and 8, q21 days). Primary end-point was progression-free survival (PFS). Results: Sixty-two patients were randomised; a total of 59 patients were treated (29 VG, 30 CG). There was no significant difference in PFS between the treatment arms: median 6.2 months for VG versus 6.3 months for CG (hazard ratio [HR]: 0.75, 95% confidence interval [CI]: 0.44–1.28; P = 0.293). Median overall survival was 12.5 months for VG versus 10.6 months for CG. The overall response rate (ORR) was higher in the VG arm than in the CG arm (63% versus 40%) but was not statistically significant in the intention-to-treat analysis. Furthermore, VG showed a high complete response (CR) rate, 22% versus 3% in CG. In the per-protocol group, both ORR and CR were significantly higher for VG than for CG. The most common adverse events (AEs) were fatigue, haematological toxicities, gastrointestinal disorders and nausea/vomiting. Common grade III/IV AEs were neutropenia (VG 62%, CG 43%), thrombocytopenia (VG 7%, CG 37%) and febrile neutropenia (VG 31%, CG 7%). Conclusions: The combination of VG did not improve PFS compared with standard treatment with CG in patients unfit for cisplatin due to renal impairment. The response rate of VG indicates, however, an active regimen and warrants further studies. Clinicaltrials.gov number: NCT02665039.