The effect of monetary contingencies on alprazolam self-administration was evaluated in seven male volunteers living in a residential laboratory. Drug administration occurred prior to an afternoon work session (13.00 h), and at the onset of an evening recreation period (17.30 h). On ‘sample’ days, participants were administered placebo or alprazolam (0.75 mg), and at the end of the afternoon work session, were told whether their task performance was ‘better’ or ‘worse’ than average. If they received ‘better’ feedback they earned 55, and if they received ‘worse’ feedback, they earned 15. On ‘choice’ days, participants chose to self-administer either alprazolam or placebo, with feedback occurring on two of the four choice days. Feedback was not actually linked to performance, but instead was pre-programmed. For one week, alprazolam administration was always associated with ‘better’ feedback on sample and choice days, and in the other week was associated with ‘worse’ feedback. When no feedback was delivered, alprazolam was self-administered equally often in the afternoon (57%) and evening (71%). When feedback was delivered, it significantly influenced the choice to self-administer alprazolam in the afternoon. ‘Better than average’ feedback resulted in alprazolam self-administration 57% of the time, but alprazolam self-administration decreased to 14% when it was associated with ‘worse than average’ feedback and reduced earnings. A similar pattern of effects has been reported for d-amphetamine. Thus, the self-administration of either a stimulant or a minor tranquilizer is significantly reduced when it is associated with a consequent loss of an alternative reinforcer, in this case money.