Oral benzodiazepine self-administration was examined in four adult male rhesus monkeys with histories of ethanol- and pentobarbital-reinforced behavior. Drug solutions and vehicle were concurrently available for 3-hr each day under fixed-ratio (FR) reinforcement schedules. Initially, the monkeys rejected a midazolam solution (0.1 mg/ml) after direct substitution of the drug for an 8% ethanol solution. However, midazolam self-administration was subsequently established by using a fading procedure in which increasing amounts of drug (0.0125-0.2 mg/ml) were gradually added to an 8% ethanol solution, followed by gradual reduction of the ethanol concentration to zero. Midazolam was an effective reinforcer for three of four monkeys tested, i.e., responding that was maintained by the drug solution exceeded that maintained by the drug vehicle. The fourth monkey also self-administered midazolam but drug-maintained responding was not consistently greater than vehicle-maintained responding. The responding maintained by the drug was an inverted-U-shaped or bitonic function of midazolam concentration. The midazolam intake (in milligrams per kilogram) increased as a function of increases in the drug concentration. At the higher concentrations, marked sedative intoxication was observed. There was an inverse relationship between FR size (varied from FR 8 to FR 32) and the amount of drug self-administered. The three monkeys in which midazolam functioned as a reinforcer were then tested with diazepam (0.2 mg/ml), which maintained drug self-administration behavior on direct substitution for 0.2-mg/ml midazolam. Diazepam-maintained responding usually exceeded water responding as the diazepam concentration was increased to 0.8 mg/ml. These data demonstrate robust reinforcing effects of both "short-" and "long-acting" benzodiazepines delivered by the oral route.