During the last decade, the Drug Monitoring allowed the clinician to control the efficiency and/or the toxicity of the used drugs. This control is of a posteriori type. The discovery of the various genetic polymorphisms (acetylation, oxidation), the use of probe drugs (caffeine, sulfadimidine, dapsone, debrisoquine, sparteine) and the fact that other drugs exhibit the same polymorphisms as the probe drugs gave the opportunity to develop non invasive methods, to study in population the distribution of the various genetic tracts, to check the subjects who might be at risk toward a drug. Can, the obtained results in simple situation be extrapolated to complex pathological or therapeutic situations? Among patients with side and/or toxic effects of the drugs, there is a significant increase of the poor metaboliser (PM) status of debrisoquine. But it is necessary to carefully check the state of the patients before phenotyping them (drug, hepatic, metabolic, renal diseases). If the patients are taking drugs which share the same genetic polymorphism as debrisoquine during the time where the phenotype is established, some misfits can happen. The patients taking propafenone present a false PM status of debrisoquine. This is less obvious with mexiletine. In a population of 200 epileptics neither the disease nor the therapeutics seem to modify the distribution of the two phenotypes PM and EM. All things considered we do not know exactly the limits of use of these tests in the patients. We need more data to validate the advices dealing with the use of pharmacogenetic concepts as a provisional tool for patient cares?