Objective To apply meta-analysis for assessing the association between ABO blood group and diabetes mellitus risk by comparing the differences of ABO blood group distribution between diabetic patients and healthy control group. Methods We searched PubMed, CBM, VIP, CNKI and Wanfang Databases to collect studies about diabetes and ABO blood group. Eligible studies were selected according to the inclusion and exclusion criteria and related data were extracted. The Newcastle-Ottawa Scale was used to assess the quality of the included studies, and meta-analysis was performed by software Review Manager 5.1 and Stata 12.0. Results Blood group B and blood group AB had a different distribution among diabetic patients and the control group. Blood group B was more frequently seen in diabetic patients (OR=1.13, 95%CI: 1.05-1.21, P=0.000 7) and blood group AB was more frequently seen in healthy control group (OR=0.86, 95%CI: 0.77-0.97, P=0.01). But only in foreign populations did subgroup analysis showed a significant difference in blood group B and blood group AB distribution among the case and control group, but not in blood group A (OR=0.90, 95%CI: 0.76-1.07, P=0.24) or the blood group O distribution (OR=1.00, 95%CI: 0.89-1.13, P=0.96). Conclusion The occurrence of diabetes mellitus may be related to ABO blood group. In foreign populations, the risk of diabetes mellitus in B blood group subjects is higher than that of other blood groups, and blood group AB may be a protective factor for diabetes. Blood group A and blood group O are not associated with diabetes in domestic or foreign populations.