Angiogenesis is a validated clinical target in advanced epithelial ovarian cancer. Cediranib is an oral antiangiogenic vascular endothelial growth factor receptor 1-3 inhibitor that has shown antitumour activity in recurrent ovarian cancer. We assessed efficacy and safety of cediranib in combination with platinum-based chemotherapy and as continued maintenance treatment in patients with first relapse of platinum-sensitive ovarian cancer. In this randomised, three-arm, double-blind, placebo-controlled phase 3 trial, we randomly assigned patients aged 18 years or older with relapsed platinum-sensitive ovarian cancer at 63 centres in Australia, Canada, New Zealand, Spain, and the UK. Participants received up to six cycles of platinum-based chemotherapy (once every 3 weeks) then entered a maintenance phase. Participants were randomly allocated (2:3:3), with five stratification factors and in alternating blocks, to receive placebo alongside chemotherapy and then placebo only maintenance (arm A; reference), cediranib 20 mg once-daily alongside chemotherapy then placebo only maintenance (arm B; concurrent), or cediranib 20 mg once-daily alongside chemotherapy then cediranib 20 mg once-daily maintenance (arm C; maintenance). Patients continued treatment to progression or excessive toxic effects. The primary efficacy endpoint was progression-free survival between arms A and C. Efficacy analysis was by intention to treat. Safety was assessed in all patients who received the allocated study drug. This trial is registered with ClinicalTrials.gov, number NCT00532194; the ISRCTN registry, number ISRCTN68510403; and ANZ Clinical Trials Registry, number ACTRN1261000016003. We randomly assigned 486 [corrected] women between Nov 13, 2007, and Dec 23, 2011; results presented are for 456 patients randomly assigned subsequent to the 30mg safety phase. During a median of 19·5 months (IQR 14-26) follow-up, 113 (96%) of 118 women assigned to arm A and 141 (86%) of 164 assigned to arm C had disease progression. Median progression-free survival was 11·0 months (95% CI 10·4-11·7) in arm C and 8·7 months (7·7-9·4) in arm A (hazard ratio 0·56, 0·44-0·72, p<0·0001). 156 (90%) of 174 patients in arm B had disease progression, and median progression-free survival was 9·9 months (95% CI 9·4-10·5). Diarrhoea, neutropenia, hypertension, and voice changes were significantly more common, during chemotherapy with cediranib, and diarrhoea, hypothyroidism and voice changes were more common during maintenance. Poor compliance with cediranib was noted during maintenance treatment with toxic effects being the most common cause for discontinuation. Cediranib, when given orally with chemotherapy and continued as maintenance, yielded a meaningful improvement [corrected] in progression-free survival in women with recurrent platinum-sensitive ovarian cancer, albeit with added toxic effects. The positive results in ICON6 could provide women with a new therapeutic option for recurrent ovarian cancer. Assessment of the secondary endpoint of overall survival will need longer follow-up. Medical Research Council, Cancer Research UK, Canadian Cancer Society Research Institute, Cancer Australia, National Gynecological Cancer Centre, and AstraZeneca.