Rats were injected with a benzodiazepine (midazolam) and shocked after presentation of an auditory conditioned stimulus (CS). They were then tested for fear reactions (freezing) to the CS in either the original context or a 2nd context after either a short (1-day) or long (21-day) retention interval. Rats tested in the original context froze less after 1 day than rats tested after that interval in the 2nd context or rats tested after 21 days. Moreover, rats tested after the long interval in the original context froze less than rats tested after that interval in the 2nd context. Therefore, midazolam does not impair the acquisition of conditioned fear but regulates when and where that fear is expressed. These effects of midazolam were interpreted as a contextually controlled deficit in the expression of conditioned fear that is similar to that associated with latent inhibition and extinction (M. E. Bouton, 1993).