AbstractWe have been able to show that the most important 5‐HT2aantagonists and agonists, belonging to chemically diverse classes can be fitted accurately into a common pharmacophoric pattern. In this paper we make use of the developed pharmacophore models to construct a fragmental amino acid model reflecting binding properties and affinity data of the structures under study. The proposed fragment model is then superimposed onto the transmembraneous part of the 5‐HT2areceptor. For this purpose the helix coordinates from the known structure of bacteriorhodopsin serve as a matrix. New structures with high affinity for the 5‐HT2asites were designed and their biological activities were predicted on the basis of interaction energies calculated with the fragment model. The predicted data show excellent agreement with experimental binding affin