Dulaglutide and liraglutide, both glucagon-like peptide-1 (GLP-1) receptor agonists, improve glycaemic control and reduce weight in patients with type 2 diabetes. In a head-to-head trial, we compared the safety and efficacy of once-weekly dulaglutide with that of once-daily liraglutide in metformin-treated patients with uncontrolled type 2 diabetes.We did a phase 3, randomised, open-label, parallel-group study at 62 sites in nine countries between June 20, 2012, and Nov 25, 2013. Patients with inadequately controlled type 2 diabetes receiving metformin (≥1500 mg/day), aged 18 years or older, with glycated haemoglobin (HbA1c) 7·0% or greater (≥53 mmol/mol) and 10·0% or lower (≤86 mmol/mol), and body-mass index 45 kg/m(2) or lower were randomly assigned to receive once-weekly dulaglutide (1·5 mg) or once-daily liraglutide (1·8 mg). Randomisation was done according to a computer-generated random sequence with an interactive voice response system. Participants and investigators were not masked to treatment allocation. The primary outcome was non-inferiority (margin 0·4%) of dulaglutide compared with liraglutide for change in HbA1c (least-squares mean change from baseline) at 26 weeks. Safety data were collected for a further 4 weeks' follow-up. Analysis was by intention to treat. This study is registered with ClinicalTrials.gov, number NCT01624259.We randomly assigned 599 patients to receive once-weekly dulaglutide (299 patients) or once-daily liraglutide (300 patients). 269 participants in each group completed treatment at week 26. Least-squares mean reduction in HbA1c was -1·42% (SE 0·05) in the dulaglutide group and -1·36% (0·05) in the liraglutide group. Mean treatment difference in HbA1c was -0·06% (95% CI -0·19 to 0·07, pnon-inferiority0·0001) between the two groups. The most common gastrointestinal adverse events were nausea (61 [20%] in dulaglutide group vs 54 [18%] in liraglutide group), diarrhoea (36 [12%] vs 36 [12%]), dyspepsia (24 [8%] vs 18 [6%]), and vomiting (21 [7%] vs 25 [8%]), with similar rates of study or study drug discontinuation because of adverse events between the two groups (18 [6%] in each group). The hypoglycaemia rate was 0·34 (SE 1·44) and 0·52 (3·01) events per patient per year, respectively, and no severe hypoglycaemia was reported.Once-weekly dulaglutide is non-inferior to once-daily liraglutide for least-squares mean reduction in HbA1c, with a similar safety and tolerability profile.Eli Lilly and Company.