Glutathione S-transferases (GSTs) catalyse reactions between glutathione and lipophilic compounds with electrophilic centres, leading to neutralisation of toxic compounds, xenobiotics and products of oxidative stress. Controversy exists about whether GST polymorphisms (GSTM1 null/present genotype, GSTT1 null/present genotype, GSTP1 Ile105Val and GSTA1 *A/*B) represent risk factors for colorectal cancer. This meta-analysis aims to examine the associations between the above-mentioned polymorphisms and colorectal cancer risk. Forty-four studies were eligible for GSTM1 (11,998 colorectal cancer cases, 17,552 controls), 34 studies for GSTT1 (8596 cases, 13,589 controls), 19 studies for GSTP1 (5421 cases, 7671 controls) and four studies for GSTA1 polymorphism (1648 cases, 2039 controls). Pooled odds ratios (ORs) were appropriately derived from fixed-effects or random-effects models. Separate analyses were conducted on Caucasian and Chinese populations. Where appropriate, sensitivity analysis concerning the deviation of genotype frequencies in controls from the Hardy-Weinberg equilibrium was performed. GSTM1 null allele carriers exhibited increased colorectal cancer risk in Caucasian populations (pooled OR=1.150, 95% confidence interval (CI): 1.060-1.248, random effects); no significant association was detected for Chinese subjects (pooled OR=1.025, 95% CI: 0.903-1.163, fixed effects). Similarly, GSTT1 null allele carriers exhibited increased colorectal cancer risk in Caucasian populations (pooled OR=1.312, 95% CI: 1.119-1.538, random effects); the association in Chinese subjects was not significant (pooled OR=1.068, 95% CI: 0.788-1.449, random effects). Concerning GSTP1 Ile105Val no significant associations were demonstrated in either race. GSTA1 *A/*B polymorphism was not associated with colorectal cancer risk. GSTM1 and GSTT1 null genotypes confer additional risk for colorectal cancer in Caucasian populations.