Background Angiogenesis is a target in the treatment of ovarian cancer. Nintedanib, an oral triple angiokinase inhibitor of VEGF receptor, platelet-derived growth factor receptor, and fibroblast growth factor receptor, has shown activity in phase 2 trials in this setting. We investigated the combination of nintedanib with standard carboplatin and paclitaxel chemotherapy in patients with newly diagnosed advanced ovarian cancer. Methods In this double-blind phase 3 trial, chemotherapy-naive patients (aged 18 years or older) with International Federation of Gynecology and Obstetrics (FIGO) IIB-IV ovarian cancer and upfront debulking surgery were stratified by postoperative resection status, FIGO stage, and planned carboplatin dose. Patients were randomly assigned (2:1) via an interactive voice or web-based response system to receive six cycles of carboplatin (AUC 5 mg/mL per min or 6 mg/mL per min) and paclitaxel (175 mg/m(2)) in addition to either 200 mg of nintedanib (nintedanib group) or placebo (placebo group) twice daily on days 2-21 of every 3-week cycle for up to 120 weeks. Patients, investigators, and independent radiological reviewers were masked to treatment allocation. The primary endpoint was investigator-assessed progression-free survival analysed in the intention-to-treat population. This trial is registered with ClinicalTrials.gov, number NCT01015118. Findings Between Dec 9, 2009, and July 27, 2011, 1503 patients were screened and 1366 randomly assigned by nine study groups in 22 countries: 911 to the nintedanib group and 455 to the placebo group. 486 (53%) of 911 patients in the nintedanib group experienced disease progression or death compared with 266 (58%) of 455 in the placebo group. Median progression-free survival was significantly longer in the nintedanib group than in the placebo group (17.2 months [95% CI 16.6-19.9] vs 16.6 months [13.9-19.1]; hazard ratio 0.84 [95% CI 0.72-0.98]; p=0.024). The most common adverse events were gastrointestinal (diarrhoea: nintedanib group 191 [21%] of 902 grade 3 and three [<1%] grade 4 vs placebo group nine [2%] of 450 grade 3 only) and haematological (neutropenia: nintedanib group 180 [20%] grade 3 and 200 (22%) grade 4 vs placebo group 90 [20%] grade 3 and 72 [16%] grade 4; thrombocytopenia: 105 [12%] and 55 [6%] vs 21 [5%] and eight [2%]; anaemia: 108 [12%] and 13 [1%] vs 26 [6%] and five [1%]). Serious adverse events were reported in 376 (42%) of 902 patients in the nintedanib group and 155 (34%) of 450 in the placebo group. 29 (3%) of 902 patients in the nintedanib group experienced serious adverse events associated with death compared with 16 (4%) of 450 in the placebo group, including 12 (1%) in the nintedanib group and six (1%) in the placebo group with a malignant neoplasm progression classified as an adverse event by the investigator. Drug-related adverse events leading to death occurred in three patients in the nintedanib group (one without diagnosis of cause; one due to non-drug-related sepsis associated with drug-related diarrhoea and renal failure; and one due to peritonitis) and in one patient in the placebo group (cause unknown). Interpretation Nintedanib in combination with carboplatin and paclitaxel is an active first-line treatment that significantly increases progression-free survival for women with advanced ovarian cancer, but is associated with more gastrointestinal adverse events. Future studies should focus on improving patient selection and optimisation of tolerability.