The purpose of this research was to characterize drug-induced rebound cue states using a three-choice, agonist-vehicle-antagonist drug-discrimination procedure. Rats were trained to discriminate among 0.50 mg/kg amphetamine (AM), distilled water (DW), and 0.03 mg/kg haloperidol (HA) in a three-lever drug discrimination task. Time-dependent changes in cue state were assessed following single doses of AM (5 and 10 mg/kg), HA (1 mg/kg), and cocaine (30 and 40 mg/kg). Consistent with expectations derived from the results of a study that used a two-lever AM-HA discrimination task, single doses of AM produced rebound responding on the HA-appropriate lever that was dose-dependent and peaked at 24 h following administration. In addition, cocaine substituted for AM at 0.5-2 h post-injection and then produced HA-like rebound responding that peaked at 24-36 h post-administration. Contrary to expectations, however, rebound AM-like responding did not occur following HA administration. Perhaps two- and three-choice discrimination tasks differ in their ability to characterize qualitative aspects of the post-haloperidol cue state. Knowledge of the time course of drug-induced adaptive processes, measured as withdrawal in the present research, is necessary for a complete description of a drug's effects and is important in understanding the effects of repeated drug administration.