Latent inhibition (LI), that is, retarded conditioning to a stimulus following its nonreinforced pre-exposure, is impaired in some subsets of schizophrenia patients and in amphetamine-treated rats. Potentiation of LI by antipsychotic drugs (APDs) given in conditioning, under conditions that do not lead to LI in controls, is a well-established index of antipsychotic activity. Recently, we have shown that the atypical APD, clozapine, in addition disrupts LI if administered in pre-exposure, under conditions that lead to LI in controls. This study demonstrates the same behavioral profile for the atypical APD risperidone. LI was measured in a thirst-motivated conditioned emotional response procedure by comparing suppression of drinking in response to a tone previously paired with a foot shock in rats that received nonreinforced exposure to the tone prior to conditioning (pre-exposed (PE)) and rats for whom the tone was novel (non-pre-exposed (NPE)). We show that under conditions that did not yield LI in vehicle controls (40 pre-exposures and five conditioning trials), risperidone (0.25, 0.5, and 1.2 mg/kg) led to LI when administered in conditioning. Under conditions that led to LI in vehicle controls (40 pre-exposures and two conditioning trials), risperidone (0.25, 0.5, and 2.5 mg/kg) abolished LI when administered in pre-exposure; the latter effect was not evident with haloperidol. In addition, the effects of risperidone administered in both the pre-exposure and conditioning stages were dose-dependent so that the pre-exposure-based action was manifested at lower but not at higher doses. It is concluded that atypical APDs exert in the LI model a dual pattern of effects, which enables detection of their 'typical' action (conditioning-based LI potentiation) as well as a dissociation from typical APDs by their 'atypical' action (pre-exposure-based LI disruption). It is suggested that the former and latter effects are subserved by D2 and 5HT2A antagonism, respectively.