Based on amino acid sequence and computer modeling, two conflicting three-dimensional models of the dopamine D2 receptor have been proposed. One model (Dahl et al., 1991, Proc. Natl. Acad. Sci. USA 88, 8111) suggests that dopamine interacts with aspartate 80 of transmembrane (TM) 2 and asparagine 390 of TM6 with the transmembranes arranged in a clockwise manner, while a second model (Hibert et al., 1991, Mol. Pharmacol. 40, 8) suggests that dopamine interacts with aspartate 114 of TM3 and the serines of TM5 (194 and 197) with the transmembranes arranged in a counterclockwise manner when viewed from the extracellular space. The present study tests the latter model by selectively mutating aspartate 114 and serines 194 and 197 of the human dopamine D2 receptor by site-directed mutagenesis. In addition, two methionines (116 and 117) were mutated to evaluate whether residues near aspartate (114) of the dopamine D2 receptor are critical in differentiating dopamine receptor agonists from adrenoceptor agonists. Removal of the negative charge with the mutation of aspartate (114) to either asparagine or glycine led to a total loss of both agonist and antagonist binding. Individual or dual methionine mutations in positions 116 and 117, to make the dopamine D2 binding pocket more closely resemble the beta 2-adrenoceptor, did not result in a change in selectivity toward noradrenergic agonists or antagonists. The serine mutations revealed interesting differences between the dopamine D2 receptor and the adrenoceptors. In particular, serine 197 appeared more important than serine 194 for agonist binding. In addition, the binding of one agonist (N-0437) was unaffected by individual serine mutations, while the binding of some antagonists, such as raclopride and spiperone, was significantly altered. These findings are discussed in relation to ligand structure and their interactions with the putative binding pocket.