Objective To perform a meta-analysis and investigate the correlation between angiotensin-I converting enzyme gene insertion (I), deletion (D) polymorphism and type 2 diabetic nephropathy, assessing the bias of small sample size study and heterogeneity between studies. Methods MEDLINE, EBSCO, EMbase, PubMed, CHKD, CNKI, CBM, VIP and WanFang Data were searched (from January 1994 to March 18th 2011) for relevant case-control studies. Two reviewers independently identified the literature according to inclusion and exclusion criteria. Also references of the included literature were retrieved. Then data were extracted and assessed by the Newcastle-Ottawa Scale (NOS). Meta-analysis was performed using RevMan 5.0.0 software. Results A total of 61 studies comprising 9 979 cases and 7 252 control subjects were included. There was b evidence of heterogeneity (Plt;0.05, I2=60%) and a random effect model was employed to summarize the data. Results of meta-analysis showed that T2DM patients with II genotype had lower incidences of DN than those with DD+DI genotype (OR=0.65, 95%CI 0.57 to 0.74). The results of subgroup meta-analysis showed that Chinese, Japanese and Brazilians patients with II genotype had lower incidences of DN. However, there were no relation among Caucasus, Middle-East, Indian, Mexican, Korean, and Malaysian patients. Conclusion As for T2DM patients, their angiotensin-I converting enzyme gene insertion (I), deletion (D) polymorphism relates to DN. T2DM patients with II genotype have lower incidences of DN. But the difference of this relation varies with ethnicity.