Sequence variation in the GSTM1, GSTT1, GSTP1, and CYP1A1 genes may potentially alter susceptibility to head and neck cancers, although evidence from previous studies has not been consistent. To explore these associations, we conducted a meta-analysis of 31 published case-control studies (4635 cases and 5770 controls) and a pooled analysis of original data from nine published and two unpublished case-control studies (2334 cases and 2766 controls). In the meta-analysis, the summary odds ratios (ORs) for head and neck cancer were 1.23 [95% confidence interval (95% CI), 1.06-1.42] for the GSTM1 null genotype, 1.17 (95% CI, 0.98-1.40) for the GSTT1 null genotype, 1.10 (95% CI, 0.92-1.31) for carrying the GSTP1 Val105 allele, and 1.35 (95% CI, 0.95-1.82) for carrying the CYP1A1 Val462 allele. The pooled analysis ORs were 1.32 (95% CI, 1.07-1.62) for the GSTM1 null genotype, 1.25 (95% CI, 1.00-1.57) for the GSTT1 null genotype, 1.15 (95% CI, 0.86-1.53) for carrying the GSTP1 Val105 allele, and 0.98 (95% CI, 0.75-1.29) for carrying the CYP1A1 Val462 allele. Increasing risk of head and neck cancer was observed with inheritance of increasing numbers of modest risk genotypes at the three GST loci (P for trend = 0.04), with the combination of carrying the GSTM1 null, GSTT1 null, and GSTP1 Val105 alleles conferring an OR of 2.06 (95% CI, 1.11-3.81). In conclusion, both the meta- and pooled analysis support modest associations of GSTM1 and GSTT1 genotypes with head and neck cancer risk, and our pooled analysis supports the notion of greater risk when genotypes at multiple GST loci are considered in a multigenic model.