Glycemic variability (GV) may be linked to the development of diabetic complications by inducing inflammation, oxidative stress, and endothelial dysfunction. Flash glucose monitoring (FGM) provides a novel method of continuously monitoring interstitial glucose levels for up to 14 days. This study randomly assigned poorly controlled type 2 diabetes mellitus patients treated with metformin and multiple daily injections of insulin (n=60) to either continuous subcutaneous insulin infusion (CSII) treatment or CSII in combination with liraglutide (CSII+Lira) treatment for 14 days during hospitalization. GV was assessed using a FGM system; weight and cardiometabolic biomarkers were also evaluated. The coefficient of variation was significantly reduced in the CSII+Lira group (P<0.001), while no significant change was observed in the CSII group. The changes differed significantly between the two groups in mean amplitude of glycemic excursions (P=0.004), standard deviation (P=0.006), and the percentage of time in the target range (4-10 mmol/L, P=0.005 and >10 mmol/L, P=0.028). The changes in mean of daily differences, interquartile range, and percentage of time in hypoglycemia (<3.3 mmol/L) and hyperglycemia (>13.9 mmol/L) identified by FGM showed no difference. Treatment with liraglutide increased serum adiponectin [33.5 (3.5, 47.7) pg/mL, P=0.003] and heme oxygenase-1 levels [0.4 (-0.0, 1.8) ng/mL, P=0.001] and reduced serum leptin levels [-2.8 (3.9) pg/mL, P<0.001]. Adding the glucagon-like peptide-1 analog liraglutide improved GV, weight, and some cardiometabolic risk markers. The FGM system is, therefore, shown to be a novel and useful method for glucose monitoring.