Caffeine metabolite ratios have been widely used to measure cytochrome P-450 1A2 activity in humans. Serum paraxanthine/caffeine ratio is one such index of this activity. We had previously demonstrated genetic variation of this trait among inbred mouse strains. In the present study, we have undertaken a genome-wide scan for quantitative trait loci affecting this trait with an interval mapping approach on an F(2) intercross population of acetaminophen nonsusceptible and C3H/HeJ inbred mice. A statistically significant association (log-likelihood ratio = 25.0) between a locus on chromosome 9, which colocalized with the murine Cyp1a2 locus, and the plasma paraxanthine/caffeine ratio was identified. This result suggested the presence of an expression polymorphism affecting this gene. A second locus was identified on chromosome 1 (log-likelihood ratio = 9.7) for which no obvious candidate gene has been identified. The influence of this locus on the paraxanthine/caffeine index was more significant among males (log-likelihood ratio = 6.3) than females (log-likelihood ratio = 3.6). A third locus was identified on chromosome 4 with a less statistically robust association (log-likelihood ratio = 3.4) to the paraxanthine/caffeine phenotype. Collectively, these three loci accounted for 63.2% of the variation observed in the F(2) population for this phenotype. These results demonstrate the potential for genetic variation arising from factors other than CYP1A2 activity to influence the plasma paraxanthine/caffeine ratio in mice. This study demonstrates the utility of quantitative genetics in the analysis of polygenic drug metabolism.