Phenotyping by probe substrates of cytochrome P450 (CYP) and other metabolizing enzymes is widely used to assess the effects of genes, environment and ethnicity on the in vivo metabolism of drugs and environmental chemicals. The caffeine metabolic ratio, in urine, plasma or saliva, has been used extensively as an index of CYP1A2, N-acetyltransferase 2 (NAT2), xanthine oxidase (XO) and CYP2A6 enzymatic activities. Phenotyping using plasma or saliva samples to measure the paraxanthine to caffeine (17X/137X) ratio correlates well with many measures of CYP1A2 activity. Various urinary metabolic ratios for caffeine phenotyping have been proposed, but shortcomings have been demonstrated for all the proposed urinary metabolic ratios. Several groups have proposed the urinary ratio of (1-methylxanthine (1X) + 1-methylurate (1U) + 5-acetylamino-6-formylamino-3-methyluracil (AFMU)) to 1, 7-dimethylurate (17U) i.e. (1X + 1U + AFMU)/17U as the preferred metabolic ratio for CYP1A2 activity (independent of urine flow rate). There is no consensus on the best urinary metabolic ratio for NAT2, XO or CYP2A6 enzymatic activities. Caffeine has been used by different groups to evaluate the in vivo activity of CYP1A2, NAT2, XO and CYP2A6 in different populations and the effect of many factors on these activities. Caffeine has been also used as a constituent of a "cocktail" to phenotype several enzymes simultaneously. In conclusion, phenotyping using caffeine as a probe substrate may still provide useful assessment of CYP1A2, NAT2, XO and CYP2A6 activities in epidemiologic and drug-drug interaction studies despite the limitations that are associated with its use.