We aimed to explore the use of the flash glucose monitoring (FGM) system in hospitalized newly diagnosed type 2 diabetes mellitus (T2DM) patients and to evaluate a new combination therapy of continuous subcutaneous insulin infusion (CSII) with or without liraglutide. This was an open-label, randomized study that was conducted in 60 newly diagnosed T2DM patients. The patients were randomized to receive either CSII (n = 30) or CSII + liraglutide (n = 30). The FGM system was used to assess the glycemic control and glycemic variability (GV) indices for 2 weeks. Mean blood glucose concentration (MBG), estimated hemoglobin A1c (HbA1c), and measures of GV, including the standard deviation of the mean glucose (SD), coefficient of variation (CV), interquartile range (IQR), mean amplitude of glycemic excursions (MAGE), largest amplitude of glycemic excursions (LAGE), and mean of daily difference (MODD) were compared between the two groups. Two oxidative stress biomarkers, 4-hydroxynonenal (4-HNE) and 8-hydroxydeoxyguanosine (8-OHdG), were measured before and after treatment. The estimated HbA1c and MBG decreased in both groups, especially the CSII + liraglutide group. SD, IQR, LAGE, and MODD were significantly lower in the CSII + liraglutide group than in the CSII group (all p < 0.05); there was no difference in CV or MAGE (p > 0.05). Similarly, the 4-HNE and 8-OHdG levels were significantly lower in the CSII + liraglutide group (p < 0.05). Our findings suggest that CSII with liraglutide was superior to CSII monotherapy in improving glycemic control and glycemic variability and in decreasing oxidative stress markers. Flash glucose monitoring can successfully provide ambulatory glucose profile data in the real world.