To investigate the association of glutathione-S-transferase (GST) polymorphisms with the risk of acute myeloid leukemia (AML), a meta-analysis of case-control studies published between 1998 and 2009 was performed. Pooled odds ratios (ORs) were assessed using both fixed- and random-effects models. Heterogeneity across studies was calculated, and funnel plots were constructed to test for publication bias. Overall, the random-effects OR with GSTM1 null genotype, GSTP1 Val105 allele and GSTT1 null genotype were 1.30 (95% confidence intervals (CI) 1.04-1.62, p = 0.018), 1.03 (95% CI 0.80-1.33, p = 0.80) and 1.24 (95% CI 0.98-1.58, p = 0.06), respectively. Statistically, significant increased risk of AML was observed with GSTM1 while borderline significance was seen with GSTT1 null genotypes. However, fixed-effects model showed significant risk of AML in the presence of null genotypes of GSTM1 and GSTT1(p < 0.05). Significant heterogeneity was found between studies relating to GSTP1 (p = 0.162), however, no heterogeneity was seen in studies that evaluated GSTM1 (Q-value = 44; I(2) = 70.9; p-value < 0.01]; and GSTT1 (Q-value = 26.03; I(2) = 57.74; p-value < 0.01] polymorphisms. From the limited studies on the association of GSTP1 with risk of AML, the role of this gene cannot be ascertained fully. Significant association of these three genes with risk of AML must be evaluated further with respect to population, smoking, eating habits, ethnicity, and race.