Pigeons were trained to discriminate the IM injection of pentobarbital (5 or 10 mg/kg) from saline in a task in which 20 consecutive pecks on one of two response keys produced access to mixed grain. Pentobarbital (1.0--17.8 mg/kg) produced a dose-related increase in the percentage of the total session responses that occurred on the pentobarbital-appropriate key. The concomitant administration of bemegride (5.6--17.8 mg/kg) antagonized the discriminative control of behavior exerted by the training dose of pentobarbital. Benzodiazepines, diazepam (1.0 mg/kg) and clobazam (3.2 mg/kg), and barbiturates, methohexital (10 mg/kg), phenobarbital (56 mg/kg), and barbital (56 mg/kg), produced responding on the pentobarbital-appropriate key similar to that produced by pentobarbital. In contrast, narcotics such as morphine, ethylketazocine, cyclazocine, and SKF-10,047, at doses up to and including those that markedly suppressed response rates, produced responding predominantly on the saline-appropriate key. Similarly, the anticonvulsants, valproate, phenytoin, and ethosuximide occasioned only saline-appropriate behavior, indicating that not all anticonvulsants share discriminative stimulus effects with pentobarbital. Muscimol, a direct GABA agonist, and baclofen, a structural analogue of GABA, also failed to produce pentobarbital-appropriate responding. Ketamine, dextrorphan, and ethanol (0.3--3.2 g/kg, orally) produced intermediate levels of pentobarbital-appropriate responding, suggesting that the discriminative effects of these drugs may be somewhat like those of pentobarbital.