Abstract Background The service capacity of primary care has improved in China. General practice also takes growing responsibility in the management of type 2 diabetes mellitus, but there are concerns about the paucity of evidence of the quality of care delivered. And there is an absence of systematic quality indicators of type 2 diabetes mellitus in general practice in China. This study aimed to develop a set of type 2 diabetes mellitus quality indicators to facilitate quality measurement in general practice in China. Methods Preliminary quality indicators were generated and refined by literature review and an expert consultation meeting. Two rounds of email-based Delphi survey and a consensus meeting were carried out to identify quality indicators. Delphi questionnaires with 43 indicators were sent to 30 participants in the first round. There were 16 general practitioners and 10 community health service center leaders from primary care, 3 endocrinologists and a primary care researcher in the first round. And 27 out of the 30 participants participated in the second round. The consensus meeting was held among 9 participants to refine the indicators and a last round of rating was carried out in the meeting. The indicators were rated in terms of importance and feasibility. The agreement criteria were defined as median ≥ 7.0 and ≥ 85.0% of ratings in the 7–9 tertile for importance; median ≥ 7.0 and ≥ 65.0, 70.0, 75.0% of ratings in the 7–9 tertile for feasibility respectively in the three rounds of rating. Results After 2 rounds of Delphi survey and the consensus meeting, total 38 indicators achieved consensus for inclusion in the final set of indicators. The final set of indicators were grouped into 7 domains: access (5 indicators), monitoring (12 indicators), health counseling (7 indicators), records (2 indicators), health status (7 indicators), patient satisfaction (2 indicators) and self-management (3 indicators). Conclusions A set of 38 potential quality indicators of type 2 diabetes mellitus in general practice were identified by an iterative Delphi process in Beijing, China. Preliminary approach for measurement and data collection were described. However, the indicators still need to be validated by testing in a further study.