The discriminative stimulus effects of benzodiazepines often have been indistinguishable from those of barbiturates or other sedative/anxiolytics. However, baboons and rats trained to discriminate lorazepam did not reliably generalize to pentobarbital in previous studies, although animals comparably trained to discriminate pentobarbital reliably generalized to lorazepam. The present study investigated the generalization profile for a variety of anxiolytic, sedative and other drugs in baboons trained to discriminate oral lorazepam (1.8 mg/kg). Triazolam, alprazolam, diazepam, midazolam, bromazepam, temazepam and nordiazepam occasioned >80% of total responses on the lorazepam-paired lever, in that order of potency, 60 min after oral dosing; chlordiazepoxide did so in three of five baboons. However, barbiturates (amobarbital, hexobarbital, methohexital, pentobarbital, phenobarbital, secobarbital) and methypryIon occasioned lorazepam-appropriate responding in only one or two baboons. Testing barbiturates at different pretreatment times (amobarbital, hexobarbital, pentobarbital or secobarbital) or by an i.m. route of administration (methohexital, pentobarbital) did not produce an increase in generalization. Neither other classic sedatives/anxiolytics (chloral hydrate, clomethiazole, ethanol, methaqualone, meprobamate, triclofos), nor anticonvulsants (phenytoin, valproic acid), nor drugs from other pharmacological classes shared discriminative-stimulus effects with lorazepam. These results, together with those from previous studies in which lorazepam or another benzodiazepine served as the training stimulus, indicate that lorazepam training results in a more selective generalization profile with respect to sedative/anxiolytic drugs than does training with other benzodiazepines.