Objective: We investigated the effect of dapagliflozin on glomerular endothelial function and tubular injury markers along with systemic endothelial function compared to metformin. Methods: We performed a single center, randomized, open-labeled, cross-over design study in patients with drug-naive type 2 diabetes. Subjects were randomized to initial metformin or initial dapagliflozin group for 8 weeks. After 1 weeks of washout period, 8 weeks’ cross-over was followed. Urine albumin excretion (UAE), N-acetyl-β-D-glucosaminidase (NAG), nitrotyrosine, and β2 microglobulin (B2M) were measured before and after new medication. Reactive hyperemic index (RHI) was measured using reactive hyperemia peripheral arterial tonometry to assess systemic endothelial function. Subjects with overt proteinuria were excluded. Results: At the baseline, mean age was 55 ± 12 years, and mean HbA1c was 7.5 ± 0.8% (n = 21). During intervention, HbA1c was significantly decreased by both of dapagliflozin and metformin (-0.48 ± 0.59% and -0.57 ± 0.65%; all P < 0.05). However, improvement of HOMA-IR, body weight reduction and increment of serum β-hydroxybutyrate level were observed only in 8-weeks of dapagliflozin treatment (P = 0.017, < 0.001, and 0.042). There was no difference in changes in HbA1c (P = 0.520), serum creatinine (P = 0.251), systolic blood pressure (P = 0.580), UAE (P = 0.697), nitrotyrosine (P = 0.678), and RHI (P = 0.990) according to medication. However, NAG was significantly decreased by metformin compared to dapagliflozin (P = 0.006). Baseline B2M level showed a significant correlation with changes in HbA1c (ρ = 0.501, P = 0.021) and serum β-hydroxybutyrate level (ρ = 0.487, P= 0.025) in dapagliflozin treatment, which was not found in metformin treatment period. Conclusion: In conclusion, 8 weeks’ dapagloflozin treatment did not affect systemic endothelial function or UAE. Baseline tubular dysfunction might be associated with poor response of dapagliflozin and ketogenesis.