In a variety of in vivo and in vitro tests, cholecystokinin (CCK) has been shown to produce effects that would suggest a functional antagonism of dopamine. On that basis, it has been hypothesized that CCK could have antipsychotic effects. We compared the CCK agonist, A68552, to the antipsychotics haloperidol (HAL), clozapine (CLOZ) and sulpiride (SULP) in various forms of conditioned avoidance using rats, mice, and cynomolgus monkeys. In rats, HAL disrupted both acquisition of a conditioned shelf-jump avoidance response and performance of the response by previously trained animals. CLOZ and SULP were ineffective in suppressing performance by previously trained rats but blocked acquisition of the response. CLOZ disrupted avoidance responding on the first 3 of 4 consecutive days of acquisition. SULP significantly suppressed avoidance responding on the last 3 days and significantly increased escape failures on day 2. A68552 administered during acquisition failed to significantly suppress avoidance responding. In mice, both HAL and CLOZ blocked performance of two-way shuttle conditioned avoidance at doses (0.1 and 3.0 mg/kg, IP, respectively) that had no effect on escape responding. A68552 at doses up to 1.07 mg/kg IP had no effect on performance. Mice treated with A68552 during acquisition showed a mild but statistically significant suppression of avoidance and an equivalent suppression of escape responding. Cynomolgus monkeys trained in a conditioned avoidance procedure were sensitive to the disruptive effects of HAL at a dose of 0.03 mg/kg IM while A68552 was without significant effect at doses up to those producing emesis (0.214 mg/kg, IM). A68552 does not resemble either HAL or the "atypical" antipsychotics, CLOZ or SULP, in conditioned avoidance tests.