Increasing evidence has demonstrated that a transcription factor 7-like 2 (TCF7L2) polymorphism (rs7903146) is significantly associated with type 2 diabetes mellitus (T2DM); however, limited sample size and variance of ethnicity in the studies investigating this association have led to conflicting reports regarding its role. Therefore, a comprehensive meta-analysis was conducted to quantitatively assess the association between the TCF7L2 polymorphism (rs7903146) and T2DM including published case-control studies in global populations. We searched the PubMed, EMbase, CNKI, and Wanfang databases for publications that studied correlation between TCF7L2 polymorphism (rs7903146) and risk of T2DM. Thirty-six studies from 30 eligible papers were identified. After data extraction and reference quality assessment, summary odds ratio and 95% confidence intervals (95%CI) of the TCF7L2 (rs7903146) polymorphism were calculated and combined using the fixed-effect model. Hardy-Weinberg equilibrium was evaluated to determine selection bias of the control subjects. Heterogeneity among studies was examined using the Q-test and the I test. Publication bias in studies was assessed using Begg's plots and the Egger test. The results showed that the rs7903146 T allele of the TCF7L2 gene was positively correlated with an enhanced risk of T2DM in the allelic, heterozygote, homozygote, dominant, and recessive models, with odds ratios of 1.35 (T vs C, 95%CI = 1.31-1.39), 1.32 (CT vs CC, 95%CI = 1.27-1.38), 1.74 (TT vs CC, 95%CI = 1.63-1.87), 1.40 (TT+CT vs CC, 95%CI = 1.35-1.46), and 1.59 (TT vs CT+CC, 95%CI = 1.49-1.69), respectively. No obvious publication bias was observed using the Egger linear test.