Aims/hypothesis Semaglutide is a glucagon-like peptide-1 analogue in development for the treatment of type 2 diabetes. Its effects on first- and second-phase insulin secretion and other measures of beta cell function and glycaemic control were assessed. Methods In this single-centre, double-blind, placebo-controlled, parallel-group trial, conducted at the Profil Institut fur Stoffwechselforschung, Germany, 75 adult (aged 18-64 years) participants with type 2 diabetes (eligibility: HbA(1c) of 6.5-9.0% (47.5-74.9 mmol/mol); BMI 20.0-35.0 kg/m(2); and treatment with diet and exercise and/or metformin monotherapy with a dose unchanged in the 30 days prior to screening) were randomised (1: 1) to once-weekly s.c. semaglutide 1.0 mg (0.25, 0.5, 1.0 mg escalated) or placebo for 12 weeks. Co-primary endpoints were changes from baseline to end of treatment in the first (AUC(0-10) (min)) and second (AUC(10-120 min)) insulin secretion phases, as measured by the IVGTT. An arginine stimulation test (AST) and a 24 h meal stimulation test were also conducted. A graded glucose infusion test (GGIT) assessed insulin secretion rate (ISR) in treated participants and a group of untreated healthy participants. Safety endpoints were also assessed. Results In total, 37 participants received semaglutide and 38 received placebo. Following IVGTT, for insulin, both AUC(0-10min) and AUC(10-120min) were significantly increased with semaglutide (estimated treatment ratio [95% CI] 3.02 [2.53, 3.60] and 2.10 [1.86, 2.37], respectively; p < 0.0001). The 24 h meal test showed reduced fasting, postprandial and overall (AUC(0-24h)) glucose and glucagon responses with semaglutide (p < 0.0001). The AST showed that maximal insulin capacity increased following semaglutide treatment. During GGIT, semaglutide significantly increased ISR to levels similar to those in healthy participants. Semaglutide was well tolerated. Conclusions/interpretation Twelve weeks of once-weekly treatment with semaglutide significantly improved beta cell function and glycaemic control in participants with type 2 diabetes.