Liraglutide is an antidiabetic agent with cardioprotective effect. The purpose of this study is to test efficacy of liraglutide to improve diabetic cardiomyopathy in patients with diabetes mellitus type 2 (DM2) without cardiovascular disease. Patients with DM2 were randomly assigned to receive liraglutide 1.8 mg/day or placebo in this double-blind trial of 26 weeks. Primary outcome measures were LV diastolic function (early (E) and late (A) transmitral peak flow rate, E/A ratio, early deceleration peak (Edec), early peak mitral annular septal tissue velocity (Ea) and estimated LV filling pressure (E/Ea), and systolic function (stroke volume, ejection fraction, cardiac output, cardiac index and peak ejection rate) assessed with CMR. Intention-to-treat analysis of between-group differences was performed using ANCOVA. Mean estimated treatment differences (95% confidence intervals) are reported. 23 patients were randomized to liraglutide and 26 to placebo. As compared with placebo, liraglutide significantly reduced E (- 56 mL/s (- 91 to - 21)), E/A ratio (- 0.17 (- 0.27 to - 0.06)), Edec (- 0.9 mL/s  * 10 (- 1.3 to - 0.2)) and E/Ea (- 1.8 (- 3.0 to - 0.6)), without affecting A (3 mL/s (- 35 to 41)) and Ea (0.4 cm/s (- 0.9 to 1.4)). Liraglutide reduced stroke volume (- 9 mL (- 16 to - 2)) and ejection fraction (- 3% (- 6 to - 0.1)), but did not change cardiac output (- 0.4 L/min (- 0.9 to 0.2)), cardiac index (- 0.1 L/min/m (- 0.4 to 0.1)) and peak ejection rate (- 46 mL/s (- 95 to 3)). Liraglutide reduced early LV diastolic filling and LV filling pressure, thereby unloading the left ventricle. LV systolic function reduced and remained within normal range. Future studies are needed to investigate if liraglutide-induced left ventricular unloading slows progression of diabetic cardiomyopathy into symptomatic stages. Trial registration ClinicalTrials.gov: NCT01761318.