AIM: To compare safety and efficacy of insulin glargine and liraglutide in patients with type 2 diabetes (T2DM). METHODS: This randomized, multinational, open-label trial included subjects treated for T2DM with metformin ± sulphonylurea, who had glycated haemoglobin (HbA1c) levels of 7.5-12%. Subjects were assigned to 24 weeks of insulin glargine, titrated to target fasting plasma glucose of 4.0-5.5 mmol/L or liraglutide, escalated to the highest approved clinical dose of 1.8 mg daily. The trial was powered to detect superiority of glargine over liraglutide in percentage of people reaching HbA1c <7%. RESULTS: The mean [standard deviation (s.d.)] age of the participants was 57 (9) years, the duration of diabetes was 9 (6) years, body mass index was 31.9 (4.2) kg/m(2) and HbA1c level was 9.0 (1.1)%. Equal numbers (n = 489) were allocated to glargine and liraglutide. Similar numbers of subjects in both groups attained an HbA1c level of <7% (48.4 vs. 45.9%); therefore, superiority of glargine over liraglutide was not observed (p = 0.44). Subjects treated with glargine had greater reductions of HbA1c [-1.94% (0.05) and -1.79% (0.05); p = 0.019] and fasting plasma glucose [6.2 (1.6) and 7.9 (2.2) mmol/L; p < 0.001] than those receiving liraglutide. The liraglutide group reported a greater number of gastrointestinal treatment-emergent adverse events (p < 0.001). The mean (s.d.) weight change was +2.0 (4.0) kg for glargine and -3.0 (3.6) kg for liraglutide (p < 0.001). Symptomatic hypoglycaemia was more common with glargine (p < 0.001). A greater number of subjects in the liraglutide arm withdrew as a result of adverse events (p < 0.001). CONCLUSION: Adding either insulin glargine or liraglutide to subjects with poorly controlled T2DM reduces HbA1c substantially, with nearly half of subjects reaching target levels of 7%.