The purposes of this study were to determine: first, if animals could be trained to discriminate a serotonin (5-HT)2 receptor agonist from a 5-HT2 receptor antagonist; second, which 5-HT2 receptor subtype was mediating the cues; and third, the usefulness of this model for studying adaptive changes in the 5-HT2 receptor system. Rats were trained to discriminate quipazine (0.5 mg/kg) from ketanserin (1.0 mg/kg) on a variable interval-20 schedule of reinforcement. After acquisition, the quipazine and ketanserin dose-response curves were found to be orderly and reproducible. Additional 5-HT2 receptor agonists (2,5-dimethoxy-4-iodoamphetamine and MK 212) and antagonists (pizotifen, mianserin, pirenperone and MDL 100,907) were tested for generalization and found to substitute for the quipazine and ketanserin cues, respectively. In antagonist studies, MDL 100,907 potently blocked quipazine discrimination. Results of ex vivo binding studies designed to estimate occupancy of 5-HT2A and 5-HT2C receptors suggested that the training dose of ketanserin blocked only 5-HT2A receptors and not 5-HT2C receptors. The combined results from the substitution, antagonism and ex vivo receptor autoradiographic studies suggest that the discriminative stimuli of quipazine and ketanserin are mediated at least in part by the 5-HT2A receptor. Additional experiments were designed to study adaptive changes in the 5-HT2A receptor. A single large dose of quipazine produced a rebound ketanserin-like effect at 20 hr after administration; however, a single large dose of ketanserin (10 mg/kg) did not produce a rebound quipazine-like effect.