Non steroidal anti-inflammatory drugs (NSAIDs) contain a chiral carbon alpha to carboxyl function. Except for naproxen, chiral NSAIDs are marketed for clinical use as racemate, ie an equimolar mixture of the two enantiomers R(-) and S(+). However, in vitro studies have shown that the anti-inflammatory activity exists almost solely in the S form. The unbound fraction is able to diffuse into tissues and to reach sites of action. It represents also the pharmacological active form. Stereoselective protein binding studies carried out at various concentrations of NSAIDs and albumin are used to evaluate the free fraction of the active enantiomer. Two optical isomers do not interact in the same manner with proteins and this binding stereoselectivity depends on NSAID and experimental conditions. Thus, it seems difficult to predict the in vivo free concentration of each enantiomer and protein binding experiments should be achieved taking into account the physiopathological parameters which influence this biological process. This enantioselectivity is determinant for the pharmacokinetic properties and could be responsible of the parameters variation obtained for each enantiomer. It could explain the variability in response to NSAIDs too. In fact, the anti-inflammatory effect is directly function of the free concentration of the S isomer. To correlate the NSAID dose with its activity, it should be better to determine this free fraction in the site of action, in particular in the synovial fluid. But the clinical response, as for example the antalgic effect, remains very far from the pharmacological activity, ie the cyclooxygenase inhibition.