The opiate antagonist, naloxone, has been reported to attenuate the loss of behavioural inhibition produced by benzodiazepines in tasks involving punishment and nonreward. This has led to suggestions that endogenous opioid systems may be involved in the anxiolytic actions of the benzodiazepines. However, the capacity of naloxone to block the effects of benzodiazepines depends on the behavioural schedule used. We tested the effects of naloxone and chlordiazepoxide on acquisition and performance of a signalled punishment schedule. Chlordiazepoxide (5 mg/kg IP) increased both punished and unpunished responding during acquisition and unpunished responding during performance of the conflict schedule. Naloxone (3 mg/kg IP) was essentially without effect on responding and failed to attenuate the punishment-releasing effects of chlordiazepoxide. The failure of naloxone and chlordiazepoxide to interact during acquisition of this punishment schedule is similar to results we obtained with successive discrimination and is in contrast to our findings with a differential reinforcement of low rates schedule. These results are consistent with the view that benzodiazepines reduce behavioural inhibition through two separate routes; that one of these routes depends on the release of endogenous opiates; and that the predominant route depends on schedule parameters.