Aims: To compare glucose control and safety of different basal insulin therapies (BI, including Insulin NPH, glargine and detemir) in real-world clinical settings based on a large-scale registry study. Methods: In this multi-center 6-month prospective observational study, patients with type 2 diabetes (HbA1c >= 7%) who were uncontrolled by oral anti-diabetic drugs (OADs) and were willing to initiate BI therapy were enrolled from 209 hospitals within 8 regions of China. Type and dose of BI were at the physician's discretion and the patients' willingness. Interviews were conducted at 0 months (visit 1), 3 months (visit 2) and 6 months (visit 3). Outcomes included change in HbA1c, hypoglycemia rate and body weight from baseline at 6 months. Results: A total of 16 341 and 9002 subjects were involved in Intention-To-Treat (ITT) and per-protocol (PP) analysis, respectively. After PS regression adjustment, ITT analysis showed that reduction in HbA1c in glargine (2.2% +/- 2.1%) and detemir groups (2.2% +/- 2.1%) was higher than that in the NPH group (2.0% +/- 2.2%) (P < .01). The detemir group had the lowest weight gain (-0.1 +/- 2.9 kg) compared with the glargine (+0.1 +/- 3.0 kg) and NPH (+0.3 +/- 3.1 kg) groups (P < .05). The glargine group had the lowest rate of minor hypoglycaemia, while there was no difference in severe hypoglycaemia among the 3 groups. The results observed in PP analyses were consistent with those in ITT analysis. Conclusion: In a real-world clinical setting in China, treatment with long-acting insulin analogues was associated with better glycaemic control, as well as less hypoglycaemia and weight gain than treatment with NPH insulin in type 2 diabetes patients. However, the clinical relevance of these observations must be interpreted with caution.