The effects of benzodiazepine (BZ) antagonists on operant behavior were examined in rats chronically administered a low dose of diazepam (DZ). The low maintenance dose of DZ (5 mg/kg twice daily) was selected as more closely associated with its anxiolytic effects than the higher treatment doses previously used to study BZ dependence. Food-restricted rats were trained to press a lever for food reinforcement under a FR20 schedule of reinforcement prior to the start of DZ administration. Acute administration of DZ caused a dose-dependent reduction of response rates, with 5 mg/kg causing a 50% decrease. Rats treated chronically with DZ became tolerant to its rate-suppressant effects as shown by a 5-fold increase in the dose of DZ required to reduce FR20 response rates by 50%. The BZ antagonist flumazenil (formerly Ro 15-1788; 10-56 mg/kg) suppressed rates of responding in rats treated chronically with DZ. The suppression of operant responding was obtained when flumazenil was given up to 3 h, but not 18 h, after the last treatment with DZ. In contrast, only the highest dose of flumazenil (56 mg/kg) caused reductions of operant responding when given to rats treated with saline. The BZ antagonist CGS 8216 (3.3-33 mg/kg IP), given 10 min prior to the session, was similarly more potent and effective at suppressing operant responding in rats treated chronically with DZ than saline. This procedure may provide a model for the clinical problem of physical dependence to chronically-administered low, anxiolytic doses of BZ tranquilizers.