Several environmental risk factors and some allelic variants of polymorphic drug-metabolising enzymes have been associated with sporadic Parkinson’s disease. No study has to date explored the possible interaction between individual susceptibility and exposure to chemical pollutants. We genotyped enzymes expressed by the human brain and involved in oxidative stress through the generation of free radicals (phase I enzymes) or involved in scavenging (phase II enzymes). Candidate genes were characterised by a genetic polymorphism that leads to lessened or abolished enzyme activity, and by fixed expression or homozygous allelic deletion (a theoretical correspondence between genotype and phenotype). Cytochrome P-450 2D6 (CYP2D6) is a non-inducible phase I enzyme involved in the biotransformation of various chemicals, including tetrahydroisoquinolines. Alleles CYP2D6*3 and CYP2D6*4 account for about 90% of the poor metaboliser autosomal recessive disorder, which has been associated with Parkinson’s disease. A meta-analysis of available studies showed an overall risk of borderline significance. Glutathione S-transferases (GSTs) are inducible phase II enzymes involved in the scavenging of many electrophilic reactive intermediates. Homozygous deletion of GSTM1 and GSTT1 loci affects, respectively, about 50% and 25% of white people (genotypes GSTM1*0 and GSTT1*0). We recruited 100 consecutive outpatients at the Institute of Neurology, University of Parma, with Parkinson’s disease (59 men, 41 women) aged 66·6 (SD 9·7) years, fulfilling the diagnostic criteria established by the UK Parkinson’s Disease Society Brain Bank. The mean age at the onset of Parkinson’s disease was 58·6 (9·7) years (mean duration of the disease 7·9 [4·6] years) and 15% of patients had a positive family history of the disease. We enrolled 200 controls (118 men, 82 women) aged 64·2 (9·2) years from outpatient specialist centres (nephrology clinic and taken on the same day as her baby’s and tested in parallel with her previous sera remained VZV IgG antibody positive and VZV IgM antibody negative. A month later, in February 1997, at another hospital in Surrey, the day after the birth of a healthy baby girl at term, birth weight 4·2 kg, her 2 –year-old sister developed chickenpox. The mother had had chickenpox 7 years previously (aged 24 years) and this was confirmed serologically by testing her stored antenatal booking serum (16 weeks’ gestation) which was VZV IgG antibody positive and VZV IgM antibody negative. Serological tests were not done on the infant at this time. Mother and baby were discharged home 2 days after delivery, with the baby well and breastfeeding. At 16 days of age the baby developed chickenpox. On day 4 of her illness the baby was seen in hospital and started on oral acyclovir. The typical rash involved the face and head, with a few lesions on the trunk and limbs. Serum from the baby at this time was VZV IgG positive and VZV IgM negative. A swab from the chickenpox lesions yielded VZV from routine tissue culture. Again, no illness or rash occurred in the mother and a repeat serum tested in parallel with her previous serum remained VZV IgG antibody positive and VZV IgM antibody negative. Although VZV was not isolated from the first baby the diagnosis was not in doubt. The strength of IgG reactivity of the mother and baby serum samples in the assay used are shown in the table. The first mother gave a test/positive control net absorbance ratio (T/P) of 1·2 (cutoff 0·8) for both early sera while that of the baby on day 4 of illness was significantly less at 0·46 (equivocal level). The 17 day postdelivery sample of the mother showed a boosted IgG response. The T/P ratios of the second mother/baby case were almost identical for the mother’s booking serum and the baby’s blood at 2·2 and 2·1, respectively—ie, both more than twice the reactivity of the positive control. In both babies the VZV IgG antibody detected (both at 4 days of illness) was likely to be entirely maternal and not their own. It is generally accepted that passively acquired maternal antibody protects neonates from varicella even in low-birthweight infants with neonatal titres of VZV IgG antibody usually matching maternal levels. We can only speculate that the reduced IgG titre in the first baby compared with his mother—possibly a result of being bottle fed—was a factor in his apparently failed immunity and more florid rash. Cell mediated immunity (CMI) was not studied in either baby. Non-specific, non-antibody-dependent cellmediated mechanisms, such as natural killer cells, are known to have a role in controlling the extent of disease due to primary varicella but cannot necessarily prevent infection in the first place. Although our two cases were of moderate severity, and both were treated with acyclovir, a defect in CMI might have been expected to have caused serious protracted disease which did not occur. There have been two previous reports of neonatal