Despite common mechanisms of actions, glucagon-like peptide-1 receptor agonists differ in structure, pharmacokinetic profile, and clinical effects. This head-to-head trial compared semaglutide with dulaglutide in patients with inadequately controlled type 2 diabetes. This was an open-label, parallel-group, phase 3b trial done at 194 hospitals, clinical institutions or private practices in 16 countries. Eligible patients were aged 18 years or older and had type 2 diabetes with HbA 7·0-10·5% (53·0-91·0 mmol/mol) on metformin monotherapy. Patients were randomly assigned (1:1:1:1) by use of an interactive web-response system to once a week treatment with either semaglutide 0·5 mg, dulaglutide 0·75 mg, semaglutide 1·0 mg, or dulaglutide 1·5 mg subcutaneously. The primary endpoint was change from baseline in percentage HbA ; the confirmatory secondary endpoint was change in bodyweight, both at week 40. The primary analysis population included all randomly assigned patients exposed to at least one dose of trial product obtained while on treatment and before the onset of rescue medication. The safety population included all randomly assigned patients exposed to at least one dose of trial product obtained while on treatment. The trial was powered for HbA non-inferiority (margin 0·4%) and bodyweight superiority. This trial is registered with ClinicalTrials.gov, number NCT02648204. Between Jan 6, 2016, and June 22, 2016, 1201 patients were randomly assigned to treatment; of these, 301 were exposed to semaglutide 0·5 mg, 299 to dulaglutide 0·75 mg, 300 to semaglutide 1·0 mg, and 299 to dulaglutide 1·5 mg. 72 (6%) patients withdrew from the trial (22 receiving semaglutide 0·5 mg, 13 receiving dulaglutide 0·75 mg, 21 receiving semaglutide 1·0 mg, and 16 receiving dulaglutide 1·5 mg). From overall baseline mean, mean percentage HbA was reduced by 1·5 (SE 0·06) percentage points with semaglutide 0·5 mg versus 1·1 (0·05) percentage points with dulaglutide 0·75 mg (estimated treatment difference ETD -0·40 percentage points 95% CI -0·55 to -0·25; p<0·0001) and by 1·8 (0·06) percentage points with semaglutide 1·0 mg versus 1·4 (0·06) percentage points with dulaglutide 1·5 mg (ETD -0·41 percentage points -0·57 to -0·25; p<0·0001). From overall baseline mean, mean bodyweight was reduced by 4·6 kg (SE 0·28) with semaglutide 0·5 mg compared with 2·3 kg (0·27) with dulaglutide 0·75 mg (ETD -2·26 kg -3·02 to -1·51; p<0·0001) and by 6·5 kg (0·28) with semaglutide 1·0 mg compared with 3·0 kg (0·27) with dulaglutide 1·5 mg (ETD -3·55 kg -4·32 to -2·78; p<0·0001). Gastrointestinal disorders were the most frequently reported adverse event, occurring in 129 (43%) of 301 patients receiving semaglutide 0·5 mg, 133 (44%) of 300 patients receiving semaglutide 1·0 mg, 100 (33%) of 299 patients receiving dulaglutide 0·75 mg, and in 143 (48%) of 299 patients receiving dulaglutide 1·5 mg. Gastrointestinal disorders were also the most common reason for discontinuing treatment with semaglutide and dulaglutide. There were six fatalities: one in each semaglutide group and two in each dulaglutide group. At low and high doses, semaglutide was superior to dulaglutide in improving glycaemic control and reducing bodyweight, enabling a significantly greater number of patients with type 2 diabetes to achieve clinically meaningful glycaemic targets and weight loss, with a similar safety profile. Novo Nordisk.