Discriminative stimulus effects of the benzodiazepine antagonist flumazenil were studied in two rhesus monkeys receiving 3.2 mg/kg/12 h of chlordiazepoxide while discriminating between vehicle and 0.056 mg/kg of flumazenil. In a drug discrimination component responding was maintained under a FR 10 schedule of stimulus-shock termination; in a non-discrimination component responding was maintained under a FR 10 schedule of food presentation. Flumazenil and Ro 15-4513 occasioned >80% flumazenil-lever responding at doses larger than 0.032 and 0.056 mg/kg, respectively. Pentylenetetrazole, ethyl-beta-carboline-3-carboxylate (betaCCE), ketamine and spiradoline failed to substitute for flumazenil although >80% drug-lever responding was observed for two of the compounds in one monkey. Flumazenil, Ro 15-4513, pentylenetetrazole, betaCCE but not ketamine or spiradoline decreased rates of responding in the food component at doses that had little effect on rates in the stimulus-shock termination component. When chlordiazepoxide injections were discontinued and saline was administered before the session, monkeys did not respond on the flumazenil lever; when flumazenil was administered under the same conditions, monkeys responded on the flumazenil lever despite not having received chlordiazepoxide for nine days. Drug stimulus control was established with flumazenil in monkeys receiving chlordiazepoxide and substitution studies suggest that this effect of flumazenil might result from antagonist actions at benzodiazepine receptors: however, lack of withdrawal-related effects after termination of chlordiazepoxide treatment precludes validation of this procedure for studying benzodiazepine dependence.