Tricyclic antidepressants were previously shown to potently inhibit the histamine-induced activation of H2 receptors linked to adenylate cyclase in cerebral membranes, and this effect was postulated to represent the mechanism of their therapeutic activity. However, these drugs were found to be much less potent (up to more than hundred-fold) at H2 receptors mediating cyclic AMP responses in intact cells. We have presently assessed whether this large difference in potency of antidepressants, also observed with haloperidol, results from a modified pharmacological specificity of the H2 receptor according to the medium composition. We have studied the binding of [125I]iodoaminopotentidine (125I-APT) to striatal or hippocampal membranes under various experimental conditions. At equilibrium the Kd of 125I-APT, a highly selective ligand for H2 receptors, was six times higher in a supplemented Tris buffer used for adenylate cyclase assays in cell-free systems than in a Krebs-Ringer medium used in studies with intact cells. The medium composition also variously affected the Ki values of the four compounds studied amitriptyline, mianserin, haloperidol and tiotidine. Whereas the Ki value of amitriptyline was little affected, that of the other compounds was four to five times lower in the supplemented Tris buffer than in the Krebs-Ringer medium. With the exception of tiotidine, the Ki values of other compounds in the binding test performed in this medium, were intermediate between those derived from the antagonism of histamine-induced cyclic AMP responses in membranes and intact cells. These data indicate that the difference between the two test responses is diversely attributable to several factors according to the compounds, the main ones being medium composition and possibly cell disruption.