In a two-lever, food-reinforced drug-discrimination paradigm separate groups of rats were trained to discriminate either arecoline, pilocarpine or oxotremorine from saline. The discriminative cues of all three agonists were potently blocked by scopolamine, but only by 30-60 fold higher doses of methylscopolamine. The three agonists all suppressed overall response rate. These rate-suppressant effects were not blocked by scopolamine in doses which blocked the discriminative cues. In generalization tests, arecoline elicited selection of the drug-appropriate lever in all groups of trained animals. Pilocarpine was discriminated as drug by all pilocarpine-trained animals and by a majority of oxotremorine-trained animals, but was not significantly discriminated by the arecoline-trained group. Oxotremorine was discriminated by all oxotremorine-trained animals but only by some pilocarpine-trained animals, and was not significantly discriminated by the arecoline-trained group. Morphine, haloperidol, chlordiazepoxide, pentobarbital and nicotine were not generalized to any of the training drugs. The discriminative stimuli produced by the training drugs are therefore specific and exhibit properties indicative of an origin at central muscarinic receptors but may not be identical.