Induction or inhibition of biotransformation enzymes, enzymes that activate or detoxify numerous xenobiotics, is one mechanism by which vegetables may alter cancer risk. Using a randomized crossover design, we examined the effect of various vegetable diets on cytochrome P450 (CYP) 1A2, N-acetyltransferase 2 (NAT2) and xanthine oxidase activity in humans. Men and women, non-smokers, on no medication and 20-40 years of age ate four 6-day controlled diets: basal (vegetable-free) and basal with three botanically defined vegetable groups. Enzyme activities were determined by measuring urinary caffeine metabolite ratios after a 200 mg caffeine dose on the last day of each feeding period. Mean CYP1A2 activity for 19 men and 17 women (least squares means adjusted for sex, GSTM1 genotype, urine volume and feeding period) with basal, brassica, allium and apiaceous vegetable diets differed significantly (P </=ISOdia</= 0. 0005) by diet, irrespective of the caffeine metabolite molar ratio used to describe CYP1A2 activity; brassica vegetables increased (P <0.04) and apiaceous vegetables decreased (P </=ISOdia</= 0.02) activity compared with the basal and allium diets. There was no effect of diet on NAT2 and xanthine oxidase activities and none of the subjects differed by GSTM1 genotype. These results demonstrate that while one vegetable subgroup induces human CYP1A2 activity, another subgroup inhibits it. This points to a complex association between consumption of a typical diet of various vegetables and biotransformation enzyme activities in humans, an association that may be difficult to interpret in observational studies.