Aims Corticosteroids are the treatment of choice for many inflammatory diseases, but often lead to adverse effects, including hyperglycemia. This study investigated the mechanisms driving differential effects on glucose control for AZD9567, a novel oral, non-steroidal, selective glucocorticoid receptor modulator, versus prednisolone in 46 patients with type 2 diabetes mellitus. Methods In this randomized, double-blind, 2-way cross-over study (NCT04556760), participants received either AZD9567 72 mg and prednisolone 40 mg daily (Cohort 1); AZD9567 40 mg and prednisolone 20 mg daily (Cohort 2); or placebo and prednisolone 5 mg daily (Cohort 3). Treatment duration was 3 days with a 3-week washout between treatment periods. Glycemic control was assessed after a standardized meal and with continuous glucose monitoring. Results A significant difference between AZD9567 and prednisolone in favour of AZD9567 was observed for the change from baseline to Day 4 glucose excursions post-meal in Cohort 1 (glucose AUC0-4h -4.54%; 95% CI: -8.88, -0.01; p=0.049), but not in Cohort 2 (-5.77%; 95% CI: -20.92, 12.29; p=0.435). In Cohort 1, significant differences between AZD9567 and prednisolone were also seen for the change from baseline to Day 4 in insulin and glucagon secretion post-meal (p<0.001 and p=0.005, respectively), and change from baseline to Day 4 in GLP-1 response (p=0.022). Significant differences between AZD9567 and prednisolone for 24-hour glucose control were observed for both Cohort 1 (-1.507 mmol/L; 95% CI: -2.0820, -0.9314; p<0.001), and Cohort 2 (-1.110 mmol/L; 95% CI -1.7257, -0.4941; p<0.001). Conclusions AZD9567 significantly reduced treatment-induced hyperglycemia compared with prednisolone.