Conflicting results in previous case-control studies on the association between Glutathione S-transferase T1 (GSTT1) gene polymorphism and Parkinson's disease (PD) risk have been reported, so we conducted this meta-analysis. We searched and extracted data from 3 Chinese and 3 English web-based electronic databases to evaluate the associations by odds ratio (OR) and its 95% confidence interval (CI) under the recessive genetic comparison model (null genotype vs. present genotype). We also conducted subgroup analyses by ethnicity and adjusted status of OR, respectively. Meta-analyses and subgroup analyses of larger studies (sample size ≥300) were also reanalyzed. When 18 eligible studies (3,963 PD cases and 5,472 controls) were pooled to analyze the association, we found no statistically significant result (OR 1.24, 95% CI 0.96-1.60). In the subgroup analyses by ethnicity, there was statistically significant association between the null genotype of GSTT1 and PD risk among Caucasians, while the associations were not found among Asians and Latinos. In the subgroup analyses by adjusted status of OR, there were no significant associations both in studies with crude OR and adjusted OR. Meta-analyses and subgroup analyses of larger studies (sample size ≥300) were also confirmed the associations mentioned above. Power analysis indicated only meta-analysis of Caucasians had enough evidence to claim the association. In conclusion, the meta-analysis suggests that the null genotype of GSTT1 contributes to PD risk in Caucasians, and no association in Asians is needed more studies to confirm.