Cytochrome P4501A2 (CYP1A2) and N-acetyltransferase-2 (NAT2) are hepatic enzymes that may activate some procarcinogens. Previous reports have determined CYP1A2 and NAT2 phenotypes by quantitating relative amounts of urinary caffeine and metabolites. However, a number of experimental issues with this approach remain. To address these, we measured caffeine and 4 metabolites in urine samples from 20 healthy volunteers on 3 separate occasions at 7-day intervals. Two additional volunteers were studied to measure the pattern of excretion of these analytes in urine over time. The molar ratio of two compounds (1,7-dimethylxanthine/1,3,7-trimethylxanthine) was used to phenotype CYP1A2, while the molar ratio of two other compounds (5-acetylamino-6-formylamino-3-methyluracil/1-methylxanthine) served to phenotype NAT2. Within-subject variation was less than 25% for most participants. In instances when within-subject variation of the metabolic ratio was > 25%, metabolite peaks were usually present in one or more control urine samples. Some caffeine metabolites were observed in urine samples at detectable levels up to 48 h after caffeine ingestion. We conclude that: (a) this assay for determining CYP1A2 and NAT2 activities (phenotyping) has an acceptably low within-subject variation over 3 consecutive weeks for most subjects who were caffeine-free for 36 h prior to study; (b) collecting and analyzing urine samples prior to testing can indicate if subjects are excreting caffeine metabolites and will aid in locating metabolite peaks on chromatograms; (c) refraining from caffeine for 48 h before testing is the best compromise between convenience for the subject and obtaining reproducible results; (d) determining metabolite molar ratios in urine collected 4-5 h after ingesting caffeine provides an acceptable time for the measurement; and (e) different ratios of metabolites for determining CYP1A2 phenotype have different advantages.