The pharmaceutical industry is committed to marketing safer drugs with fewer side effects, predictable pharmacokinetic properties and quantifiable drug-drug interactions. Drug metabolism is a major determinant of drug clearance and interindividual pharmacokinetic differences, and an indirect determinant of the clinical efficacy and toxicity of drugs. Progressive advances in the knowledge of metabolic routes and enzymes responsible for drug biotransformation have contributed to understanding the great metabolic variations existing in human beings. Phenotypic as well genotypic differences in the expression of the enzymes involved in drug metabolism are the main causes of this variability. However, only a minor part of phenotypic variability in man is attributable to gene polymorphisms, thus making the definition of a normal liver complex. At present, the use of human in vitro hepatic models at early preclinical stages means that the process of selecting drug candidates is becoming much more rational. Cultured human hepatocytes are considered to be the closest model to human liver. However, the fact that hepatocytes are located in a microenvironment that differs from that of the cell in the liver raises the question: to what extent does drug metabolism variability observed in vitro actually reflect that of the liver in vivo? By comparing the metabolism of a model compound both in vitro and in vivo in the same individual, a good correlation between the in vitro and in vivo relative abundance of oxidized metabolites and the hydrolysis of the compound was observed. Thus, it is reasonable to consider that the variability observed in human hepatocytes reflects the existing phenotypic heterogeneity of the P450 expression in human liver.