The correct in vivo quantification of aryl hydrocarbon receptor-mediated induction of cytochrome P450 1A2 (CYP1A2) in humans is a long-standing question. We compared the performance of several modifications of the caffeine test for measurement of CYP1A2 activity in subjects with exceptionally high, low, or absent enzyme induction. CYP1A2 activity was measured in 2 women highly exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), in 1 man moderately exposed, and in 50 control subjects (30 nonsmokers and 20 heavy smokers). After the application of a test dose, caffeine demethylation was detected with the carbon 13 breath test, the total clearance, and several serum and urinary metabolite ratios. In the highly TCDD-exposed persons, results of the breath test (cumulative 15-minute dose), the total caffeine clearance, the serum metabolic ratio paraxanthine/caffeine (30 and 120 minutes after application), and the urinary metabolic ratio sum of 5-acetylamino-6-formylamino-3-methyluracil (AFMU), 1-methyluric acid (1U), and 1-methylxanthine (1X) over 1,7-dimethyluric acid (17U) showed a CYP1A2 activity 8 to 10 times higher than the mean of nonsmokers. In contrast, two caffeine urinary metabolic ratios with the parent substance in the denominator did not reflect the CYP1A2 enzyme induction. These ratios strongly depended on urine flow. For the breath test, only results evaluated for a short sampling period (eg, 15 minutes after application) revealed the high induction. Compared with nonsmokers, higher mean values (maximally 1.8 times) were observed in smokers with all tests. After high TCDD exposure, hepatic CYP1A2 activity is inducible at least 10 times in humans. Moderate TCDD exposure (up to 1000 ppt in blood fat) does not cause a CYP1A2 induction that can be measured to differentiate from background exposure individually. Therefore direct quantification of such toxins is more specific and sensitive.