Aim To determine equivalence of modified gemcitabine and oxaliplatin compared with gemcitabine and cisplatin in unresectable gallbladder cancer (GBC). Primary end-point was overall survival (OS). Methods Open label, prospective, randomised phase III equivalence study. Inclusion criteria included histologically proven unresectable GBC, 18 years or older, adequate organ functions and Eastern Cooperative Oncology Group ≤2. Sample size 108 patients were required in each arm to have an equivalence margin of ±2 months with power of 80%. Treatment Modified gemcitabine and oxaliplatin (mGemOx)—gemcitabine 900 mg/m2, oxaliplatin 80 mg/m2, maximum 6 cycles; gemcitabine + cisplatin (CisGem)—gemcitabine 1000 mg/m2, cisplatin 25 mg/m2, maximum 8 cycles, all day 1 and 8 every 3 weeks. Results Two hundred sixty subjects were recruited between February 2011 and July 2015. Two hundred forty-three patients (119, mGemOx and 124, CisGem) received at least 1 dose and analysed for safety and efficacy (modified intention to treat). Median OS was 8·5 months for whole group (95% confidence interval [CI]: 7·9–9·1). Median OS in mGemOx was 9 months and 8·3 months in CisGem; p = 0·057 (hazard ratio = 0·78; 95% CI = 0·60-1·02). Restricted mean OS for follow-up limited to 30 months was 11·2 months (95% CI: 9·8–12·6) in mGemOx and 10·4 months (95% CI: 9·1–11·7) in CisGem. Difference of the mean was 0·8 months with 95% CI, exceeding 2 months (−1·1 to 2·7), hence rejecting equivalence. Peripheral neuropathy, thrombocytopaenia in mGemOx and nephrotoxicity was higher with CisGem. Conclusion This trial failed to show equivalence of eight cycles of CisGem to six cycles of mGemOx. Numerically OS was better with mGemOx. Toxicities were different. The trial was not powered to answer superiority. Clinical trial registration CTRI/2010/091/001406.