This study aimed to evaluate endogenous metabolic markers of hepatic cytochrome P450 (CYP)3A activity in healthy subjects using a metabolomics approach. Twenty‐four subjects received the following medication during the following three study periods: 1 mg of i.v. midazolam alone (control phase), 1 mg of i.v. midazolam after 4 days of pretreatment with 400 mg of ketoconazole once daily (CYP3A‐inhibited phase), and 2.5 mg of i.v. midazolam after 10 days of pretreatment with 600 mg of rifampicin once daily (CYP3A‐induced phase). During each study period, 24 h before and after the administration of midazolam, urine samples were collected at 12‐h intervals for metabolomic analyses. We derived an equation to predict midazolam clearance (CL) based on several of these markers. We demonstrated that a combination of the concentrations and ratios of several endogenous metabolites and the CYP3A5*3 genotype is a reliable predictive marker of hepatic CYP3A activity as assessed by i.v. administration of midazolam. Clinical Pharmacology & Therapeutics (2013); 94 5, 601–609. doi:10.1038/clpt.2013.128