A new pharmacophoric model for the H1-antagonist binding site is derived which reveals that a simple atom to atom matching of compounds is not sufficient; in this model, interacting residues from the receptor need to be included. To obtain this model, the bioactive conformations of several (semi-)rigid classical histamine H1-receptor antagonists have been investigated (cyproheptadine, phenindamine, triprolidine, epinastine, mequitazine, IBF28145, and mianserine). In general, these antihistamines contain two aromatic rings and a basic nitrogen atom. A previously derived pharmacophoric model with the nitrogen position fixed relative to the two aromatic rings is now found not to be suitable for describing the H1-antagonist binding site. A procedure is described which allows for significant freedom in the position of the basic nitrogen of the histamine H1-antagonist. The area accessible to the basic nitrogen is confined to the region accessible to its counterion on the histamine H1-receptor, i.e., the carboxylate group of Asp116. The basic nitrogen is assumed to form an ionic hydrogen bond with this aspartic acid which C alpha- and C beta-carbons are fixed with respect to the protein backbone. Via this hydrogen bond, the direction of the acidic proton of the antagonist is taken into account. Within these computational procedures, an aspartic acid is coupled to the basic nitrogen of each H1-antagonist considered; the carboxylate group is connected to the positively charged nitrogen via geometric H-bonding restraints obtained from a thorough database search (CSD). Also to the basic nitrogen of the pharmacophore is coupled an aspartic acid (to yield our new template). In order to derive a model for the H1-antagonist binding site, the aromatic ring systems of the antagonists and template are matched according to a previously described procedure. Subsequently, the C alpha- and C beta-carbons of the aspartic acid coupled to the H1-antagonists are matched with those of the template in a procedure which allows the antagonist and the carboxylate group to adapt their conformation (and also their relative position) in order to optimize the overlap with the template. A six-point pharmacophoric model is derived which has stereoselective features and is furthermore able to distinguish between the so-called "cis"- and "trans"-rings mentioned in many (Q)SAR studies on H1-antagonists. Due to its stereoselectivity, the model is able to designate the absolute bioactive configuration of antihistamines such as phenindamine (S), epinastine (S), and IBF28145 (R). A further merit of this study is that a model is obtained which includes an amino acid from the receptor.(ABSTRACT TRUNCATED AT 400 WORDS)