The purpose of the present study was to further investigate the relationship between the DS effects of PB and those of benzodiazepines (BZs) and to begin to collect pharmacological information concerning receptor mechanisms involved in this behavioral effect of BZs. Rhesus monkeys (n = 3), trained to discriminate pentobarbital (PB; 10 mg/kg, IG) from saline under a discrete-trials shock avoidance procedure, were given IG diazepam (0.3-10 mg/kg), chlordiazepoxide (1.0-30 mg/kg), or etizolam (0.3-10 mg/kg) alone and in combination with flumazenil (0.01-1.7 mg/kg, IM). Flumazenil was administered 10 min prior to the administration of saline, PB or the BZs. All three BZs fully substituted for PB in all monkeys. Diazepam was the most potent with a mean ED50 of 0.81 mg/kg (SEM = 0.04) while chlordiazepoxide was the least potent (mean ED50 = 5.78 mg/kg, SEM = 1.22 mg/kg). The ED50 for etizolam was 1.22 mg/kg (SEM = 0.37 mg/kg). Pretreatment with flumazenil (0.01-1.0 mg/kg) resulted in a dose-related parallel shift to the right in the dose-response function for PB-appropriate responding in all monkeys for all three BZs. The mean (n = 3) pKB value with 0.1 mg/kg flumazenil was 6.51 (SEM = 0.42) for diazepam and 6.57 (SEM = 0.17) for chlordiazepoxide. This value could not be calculated for etizolam because only one monkey was tested with 0.1 mg/kg flumazenil. However, the mean pKB for etizolam considering all monkeys and all doses of flumazenil was 6.58 (SEM = 0.47). Apparent pA2 values for flumazenil with diazepam were 6.02 for one monkey and 7.11 for another. All three BZs tended to increase average latency to respond. Apparent pKB and pA2 analysis may prove useful for elucidating receptor mechanisms involved in the behavioral effects of BZs.