In the treatment of inflammatory skin diseases, there are some glucocorticoid (GC) double esters combining pronounced antiinflammatory activity and minor atrophogenic side effects. The reason, however, is only poorly understood. To investigate interactions of GCs with the ligand-binding domain of the glucocorticoid receptor (GR), we measured receptor-binding potency of a series of GC esters including their metabolites and performed a molecular modeling study using progesterone receptor crystal structure data. Ligand docking to the GR-binding pocket showed good fitting of GC 17-esters corresponding to their high receptor-binding affinity, and unfavorable sterical interactions for GC 21-esters with substituents larger than propionate. Molecular dynamics simulations served to visualize induced fit procedures. Ligand docked GC conformations after dynamics simulations were used for generation of a 3D quantitative structure-activity relationship model. Using a set of 11 steroids, this model showed a correlation coefficient (r(2)) of 0.98, a leave-one-out cross validation (q(2)) of 0.79 and was able to predict binding affinity of further six ligands with a standard error of prediction of 0.33. Moreover, interactions of Asn-564 and Met-639 with the steroids were investigated by studying GR mutants of these amino acids. Met-639 participates in hydrophobic interactions mainly with GC side chains, while Asn-564 forms a hydrogen bond to the C11-OH group of the steroid. Asn-564 is shown to be very important for ligand binding and even more for target gene activation and transcription factor repression.