The present studies examined the effects of chronic posttest treatment with the antipanic agent alprazolam (ALP) or the traditional anxiolytic agents chlordiazepoxide (CDP) and phenobarbital (PhB) on conflict behavior. In daily ten-minute sessions, water-deprived rats were trained to drink from a tube which was occasionally electrified (0.25 or 0.5 mA). Electrification was signalled by a tone. Chronic ALP (10 mg/kg/day), CDP (40 mg/kg/day), PhB (80 mg/kg/day) or vehicle were injected IP after conflict testing (in some experiments again 12-16 hours later) for a minimum of 6 weeks. Chronic ALP (but not CDP or PhB) resulted in a time-dependent increase in punished responding, with a latency to onset of 3-4 weeks; this effect was not antagonized by the benzodiazepine antagonist Ro15-1788. These data support the hypothesis that conflict paradigms may serve as animal models for the study of antipanic agents. Moreover, these data suggest that not all anxiolytics will exhibit antipanic efficacy.