The suitability of the endogenous 6-hydroxymelatonin/melatonin urinary metabolic ratio as a surrogate for the paraxanthine/caffeine ratio to predict cytochrome P450 1A2 (CYP1A2) activity was assessed in this study. Twelve healthy volunteers completed four study sessions spread over 1 month (including overnight urine collection with first morning voids collected separately). Except for the third session, volunteers were asked to abstain from methylxanthine-containing beverages and foods at least 24 h before urine collection. At the end of urine collection, subjects were given a caffeinated beverage and capillary blood samples were collected 2 h after the drink administration. A significant linear relationship between the 6-hydroxymelatonin/melatonin ratios from 12-h urine samples and first morning voids was observed (R ² = 0.876, p < 0.0001). In contrast to the paraxanthine/caffeine ratio, consumption of methylxanthine-containing beverages during session three did not significantly influence the 6-hydroxymelatonin/melatonin ratios compared with the other sessions requiring abstinence from caffeine. A larger intra- and interindividual variability in the 6-hydroxymelatonin/melatonin ratios compared with the paraxanthine/caffeine ratio was also observed. A very weak correlation was observed between the paraxanthine/caffeine ratio and both of the endogenous 6-hydroxymelatonin/melatonin ratios (Pearson r < 0.35, p < 0.05). All these results question whether this endogenous metric could adequately reflect CYP1A2 activity or substitute for the probe caffeine. Additional studies with larger study samples are needed to examine this endogenous metric in more details.