The aim of this study is to assess the association of the SCGB3A2 -112G>A promoter polymorphism with Graves' disease(GD) using a meta-analysis. Relevant studies were identified using PubMed and EMBASE electronic databases. A meta-analysis of relevant studies was performed. This meta-analysis included four case-control studies, containing 6,913 GD cases (Caucasian 3904, Han 3009) and 7,185 controls(Caucasian 4155, Han 3030). The combined results showed a significant difference in genotype distribution (-112A/G) between GD and control populations (A vs. G P = 1.53 × 10(-7); GG vs. AA+AG P = 6.78 × 10(-9)). Meta-analysis was performed using a fixed-effects model. Under the dominant model (GG/AA + GA), the AA and GA genotypes were significantly associated with GD (pooled OR = 1.24, 95 % CI 1.12-1.37). When the two European studies are combined, the AA and GA genotypes were also significantly associated with GD (pooled OR = 1.29, 95 % CI 1.20-1.39). This meta-analysis suggests that SCGB3A2 polymorphism at positions -112G>A was associated with GD both in Chinese and Caucasian population.