Caffeine is sequentially metabolized by cytochrome P4501A2 (CYP1A2), N-acetyltransferase (NAT) and/or xanthine oxidase (XO). In the present study the activity of these three enzymes was estimated from ratios of the metabolites formed from dietary caffeine and excreted into the urine collected as spot samples. In the urine samples from 10 out of 377 subjects concentrations of caffeine metabolites were too low to allow reliable measurements of the ratios. In 335 healthy subjects the NAT activity showed a typically bimodal distribution with 47% fast acetylators and 53% slow acetylators, consistent with a Danish population. The ratios reflecting CYP1A2 and XO activities were log normal and normal distributed, respectively. In 103 non-smoking men and 90 non-smoking women the ratio of caffeine metabolites expressing CYP1A2 activity was 4.7 +/- 1.6 and 4.3 +/- 1.9 as compared to 7.8 +/- 2.5 and 7.3 +/- 3.0 in 31 male and 25 female subjects smoking 10 cigarettes/day or more respectively, verifying induction of CYP1A2 by tobacco (P less than 0.05), but minimal sex-related differences. In 12 non-smoking pregnant women and in 28 women using oral contraceptives the CYP1A2 ratio was 29 and 20% reduced respectively (P less than 0.05). In a multivariate analysis the only significant predictor of the XO ratio was the consumption of caffeine with an increase of 2% per cup of coffee or equivalent (P less than 0.05). In 23 healthy male subjects 30 days of vigorous exercise increased the CYP1A2 ratio by 70% and the XO ratio by 42% (P less than 0.05), but left the NAT ratio unchanged. In nine healthy volunteers daily ingestion of 500 g of broccoli for 10 days increased the CYP1A2 ratio by an average of 12% (P less than 0.05), compared to a control period with ingestion of an equivalent weight of non-cruciferous green vegetables. The ratios of metabolites from dietary caffeine in spot urine samples offer ethical, non-invasive and reliable estimates of CYP1A2, NAT and XO. These enzymes are highly relevant for the bioactivation of potentially toxic compounds and the formation of oxygen radicals. The method is applicable in large-scale epidemiological studies, allowing, for example, prospective testing of the relationship between these enzyme activities and the development of disease. Exercise may increase CYP1A2 activity to a magnitude corresponding to heavy smoking, as well as XO by mechanisms that remain to be clarified.