Aim To assess the efficacy and safety of a dipeptidyl peptidase-4 (DPP-4) inhibitor combined respectively with three oral antihyperglycaemic agents in Chinese patients with newly diagnosed type 2 diabetes mellitus (T2DM) with high levels of glycated haemoglobin (HbA1c). Materials and methods Between 30 December 2014 and 1 November 2017, a 24-week, multicentre, parallel-controlled study was performed on drug-naive T2DM patients. In total, 648 patients with 8.0% <= HbA1c <= 11.0%, aged 18-80 years and body mass index (BMI) 19-40 kg/m(2) were randomly assigned 1:1:1 to receive saxagliptin (Saxa) combined with metformin (Met), acarbose (Aca) or gliclazide (Gli) modified release (MR) tablets (Saxa + Met, Saxa + Aca and Saxa + Gli). The primary outcome was the absolute change in HbA1c from baseline; secondary outcome was the percentage of patients achieving HbA1c Each treatment arm contained 216 patients; overall, 583 completed the 24-week trial. At 24 weeks, the mean (95% confidence interval) change in HbA1c from baseline in Saxa + Met, Saxa + Aca and Saxa + Gli were, respectively: -2.9% [-3.1, -2.8]; -2.6% [-2.8, -2.5]; and -2.8% [-2.9, -2.6] (overall p = .04, Saxa + Aca vs. Saxa + Met, p = .010, Saxa + Gli vs. Saxa + Met, p = 0.18). At 24 weeks, 84.9%, 74.7% and 80.3% of participants were at HbA1c <7.0% (overall p = .05); and 72.6%, 59.8% and 63.3% were HbA1c <= 6.5% (overall p = 0.10). The rates of minor or symptomatic hypoglycaemia were very low. Conclusions Initial treatment with a DPP-4 inhibitor combined with Metform, alpha-glycosidase inhibitor or sulphonylurea was safe and effective for patients with newly diagnosed T2DM and high HbA1c. DPP-4 inhibitor combined with Met showed the best efficacy for this population.