Glutathione S-transferase GSTM1, GSTM3 and GSTT1 and cytochrome P450 CYP2D6, CYP1A1 and CYP2E1 loci are susceptibility candidates for cancers of the upper aerodigestive tract because putatively protective and risk genotypes have been identified from studies in other diseases associated with alcohol and tobacco consumption. We describe genotype frequencies in 398 oral, pharyngeal and laryngeal squamous cell carcinoma patients and 219 control individuals. Of the genotypes presumed to be protective, only GSTM1 A/B influenced susceptibility; the GSTM1 A/B frequency was lower in the patients than the control individuals both before [odds ratio = 0.3, 95% confidence interval (CI) 0.1-0.7] and after correction for imbalances in age, sex, smoking and alcohol consumption (odds ratio = 0.2, 95% CI 0.1-0.5). Of the putatively risk genotypes, GSTM3 AA, previously associated with susceptibility to skin cancer, was higher in the cases (odds ratio = 1.6, 95% CI 1.1-2.4). Dividing cases into oral/pharyngeal and laryngeal squamous cell carcinoma showed the GSTM3 AA frequency was higher in laryngeal squamous cell carcinoma than control individuals (odds ratio = 1.6, 95% CI 1.1-2.5) and the difference between control individuals and oral/pharyngeal squamous cell carcinoma approached significance (odds ratio = 1.7, 95% CI 1.0-2.8). The putatively protective GSTM3 BB genotype was lower in patients with glottic (1.0%) than supraglottic (3.0%) squamous cell carcinoma. We identified no differences between patients and control individuals in the frequencies of presumed risk genotypes (e.g. CYP2D6 EM, CYP1A1 m1/m1, CYP1A1 Ile/Ile, CYP2E1 DD, CYP2E1 c1c1, GSTT1 null) or, interactions between genotypes and smoking or alcohol consumption. We conclude, first, that mu class glutathione S-transferase influence risk of upper aerodigestive tract cancers thereby complementing studies in skin cancer patients showing GSTM1 A/B is protective, while GSTM3 AA moderately increases risk. The influence of GSTM1 A/B, but not GSTM1 A or GSTM1 B (mostly heterozygotes with GSTM1*0) suggests that two expressed alleles may attenuate risk. While we found immunohistochemical evidence of GSTM3 expression in the cilia lining the larynx, the biochemical consequences of the polymorphism are unclear. Indeed, the influence of the gene may reflect linkage disequilibrium with another gene. However, we did not find an association with GSTM1 genotypes. Second, we conclude that the CYP2D6, CYP2E1, CYP1A1 and GSTT1 alleles studied, although putatively good candidates, either do not determine the effectiveness of detoxification of tobacco-derived carcinogens in the upper aerodigestive tract or, that chronic consumption of tobacco and alcohol overwhelms enzyme defences, irrespective of genotype.