AIM: To examine the effect of combination therapy with canagliflozin plus liraglutide versus each agent alone on beta cell function in type 2 diabetes mellitus (T2DM) patients. RESEARCH DESIGN AND METHODS: A total of 45 poorly controlled (HbA1c = 7%-11%) T2DM patients received an oral glucose tolerance test (OGTT) before and after 16 weeks of treatment with: (i) liraglutide (LIRA); (ii) canagliflozin (CANA); (iii) liraglutide plus canagliflozin (CANA/LIRA). RESULTS: Both liraglutide and canagliflozin significantly lowered HbA1c with no significant additive effect of the combination on HbA1c (0.89%, 1.43%, and 1.67% respectively). Insulin secretion during the OGTT, measured with (∆C-Pep/∆G)0-120, increased in the 3 groups (from 0.30 ± 0.06 to 0.48 ± 0.10; 0.29 ± 0.05 to 0.98 ± 0.23; and 0.24 ± 0.06 to 1.09 ± 0.12 in subjects receiving CANA, LIRA and CANA/LIRA respectively; P = 0.02 for CANA vs LIRA, P < 0.0001, CANA/LIRA vs CANA), and the increase in insulin secretion was associated with an increase in beta cell glucose sensitivity (29 ± 5 to 55 ± 11; 33 ± 6 to 101 ± 16; and 28 ± 6 to 112 ± 12, respectively; P = 0.01 for CANA vs LIRA, P < 0.0001, CANA/LIRA vs CANA). No significant difference in the increase in insulin secretion or beta cell glucose sensitivity was observed between subjects in LIRA or CANA/LIRA groups. The decrease in HbA1c strongly and inversely correlated with the increase in beta cell glucose sensitivity (r = 0.71, P < 0.001). In multivariate regression model, improved beta cell glucose sensitivity was the strongest predictor of HbA1c decrease with each therapy. CONCLUSION: Improved beta cell glucose sensitivity with canagliflozin monotherapy and liraglutide monotherapy or in combination is major factor responsible for the HbA1c decrease. Canagliflozin failed to produce an additive effect to improve beta cell glucose sensitivity above that observed with liraglutide.