CONTEXT: The mechanisms of action of incretin mimetics in patients with long-standing type 2 diabetes (T2D) and high insulin requirements have not been studied. OBJECTIVE: To evaluate changes in ß-cell function, glucagon secretion, and fat distribution after addition of liraglutide to high-dose insulin. DESIGN: A single-center, randomized, double-blind, placebo-controlled trial. SETTING: University of Texas Southwestern and Parkland Memorial Hospital clinics. PATIENTS: Seventy-one patients with long-standing (median, 17 years) T2D requiring high-dose insulin treatment (>1.5 U/kg/d; average, 2.2 ± 0.9 U/kg/d). INTERVENTION: Patients were randomized to liraglutide 1.8 mg/d or matching placebo for 6 months. MAIN OUTCOME MEASURES: We measured changes in insulin and glucagon secretion using a 4-hour mixed-meal challenge test. Magnetic resonance-based techniques were used to estimate sc and visceral fat in the abdomen and ectopic fat in the liver and pancreas. RESULTS: Glycosylated hemoglobin improved significantly with liraglutide treatment, with an end-of-trial estimated treatment difference between groups of −0.9% (95% confidence interval, −1.5, −0.4%) (P = .002). Insulin secretion improved in the liraglutide group vs placebo, as measured by the area under the curve of C-peptide (P = .002) and the area under the curves ratio of C-peptide to glucose (P = .003). Insulin sensitivity (Matsuda index) and glucagon secretion did not change significantly between groups. Liver fat and sc fat decreased in the liraglutide group vs placebo (P = .0006 and P = .01, respectively), whereas neither visceral nor pancreatic fat changed significantly. CONCLUSIONS: Treatment with liraglutide significantly improved insulin secretion, even in patients with long-standing T2D requiring high-dose insulin treatment. Liraglutide also decreased liver and sc fat, but it did not alter glucagon secretion.