An isocratic high-performance liquid chromatography procedure was developed for the analysis of five urinary metabolites of caffeine; caffeine or 1,3,7-trimethylxanthine (137X), paraxanthine or 1,7-dimethylxanthine (17X), 1,7-dimethylurate (17U), 1-methylxanthine (1X), 1-methylurate (1U) and 5-acetylamino-6-formylamino-3-methyluracil (AFMU). A standardized procedure was used for oral intake of caffeine and for urine collection. Conditions for sample storage and preparation were optimized, resulting in no detectable loss of caffeine metabolites after storage of the urine samples for four months. Urine samples were extracted with chloroform-2-propanol (4:1, v/v) and separated on a reversed-phase column with acetic acid (33%)-tetrahydrofuran-acetonitrile-water (1:2.5:44:925.5, v/v) as the eluent. Peaks were monitored at 280 nm. Peak heights were measured and the five metabolites were quantified using calibration curves. Cytochrome P4501A2 (CYP1A2) activity was calculated from the molar ratio (AFMU+1X+1U)/17U, N-acetyltransferase (NAT) from the ratio AFMU/1X, XO from the ratio 1U/1X+1U and cytochrome P4502A6 (CYP2A6) from the ratio 17U/(17U+17X+1U+ 1X+AFMU). The inter-assay coefficients of variation ranged from 1.7% for 17U to 5.7% for IX. The intra-individual variation in metabolite ratios determined in two people, with intervals of a few days to several weeks between measurements, ranged from 2. 1% for XO to 11.0% for CYP2A6. Using this procedure, metabolic ratios were determined for four groups of subjects; healthy, non-smoking females using oral contraceptives (OC users, n=5) and non-users (n=5), healthy nonsmoking males (n=9) and children (n=7). Results found in this study were comparable to results reported in the literature for subjects with similar characteristics. A significantly higher CYP1A2 ratio was found for males (4.87+/-0.47) compared to females (3.62+/-0.91: p=0.005, Mann-Whitney). For the other enzyme activities, no significant differences were found between the groups of subjects in this study.