Background: Diabetic retinopathy remains a leading cause of visual loss in people of working age. We examined whether candesartan treatment could slow the progression and, secondly, induce regression of retinopathy in people with type 2 diabetes. Methods: We did a randomised, double-blind, parallel-group, placebo-controlled trial in 309 centres worldwide. We recruited normoalbuminuric, normotensive, or treated hypertensive people with type 2 diabetes with mild to moderately severe retinopathy and assigned them to candesartan 16 mg once a day or placebo. After a month, the dose was doubled to 32 mg once per day. Investigators and patients were unaware of the treatment allocation status. Progression of retinopathy was the primary endpoint, and regression was a secondary endpoint. Analysis was by intention to treat. The trial is registered with ClinicalTrials.gov, number NCT00252694. Findings: 1905 participants (aged 37-75 years) were randomised to candesartan (n=951) or placebo (n=954). 161 (17%) patients in the candesartan group and 182 (19%) in the placebo group had progression of retinopathy by three steps or more on the Early Treatment Diabetic Retinopathy Study scale. The risk of progression of retinopathy was non-significantly reduced by 13% in patients on candesartan compared with those on placebo (hazard ratio [HR] 0縷, 95% CI 0緰-1簸, p=0粰). Regression on active treatment was increased by 34% (1糴, 1簸-1綸, p=0簰9). HRs were not attenuated by adjustment for baseline risk factors or changes in blood pressure during the trial. An overall change towards less severe retinopathy by the end of the trial was observed in the candesartan group (odds 1籷, 95% CI 1簵-1糰, p=0簰3). Adverse events did not differ between the treatment groups. Interpretation: Treatment with candesartan in type 2 diabetic patients with mild to moderate retinopathy might induce improvement of retinopathy. Funding: AstraZeneca and Takeda.