It has been argued that the control of instrumental action by motivational states is indirect, being mediated by the effects these states have on the incentive value of the instrumental outcome (A. Dickinson & B. W. Balleine, 1994). In this study, the benzodiazepine agonist midazolam was found to control instrumental action in a similar manner. Midazolam (1 mg/kg) increased rewarded instrumental performance in rats trained undeprived but only elevated performance in extinction when rats were reexposed to the instrumental outcome under midazolam before the test. This effect of reexposure under midazolam did not transfer to a test conducted under food deprivation, suggesting that it was not produced by drug-induced hunger. Finally, in animals trained hungry, midazolam was found to block outcome devaluation induced by a reduction in food deprivation, indicating that midazolam and food deprivation affect outcome value via a common substrate.