Rats were trained to discriminate nicotine from saline in a two-bar operant conditioning procedure with food reinforcement. There was partial generalization to the nicotine analogues anabasine and cytisine in rats trained to discriminate either 0.2 or 0.4 mg/kg nicotine from saline. However, generalization was complete in rats trained to discriminate 0.1 mg/kg nicotine and, in a novel procedure, any one of three doses of nicotine (0.1, 0.2, or 0.4 mg/kg). There was no generalization to the muscarinic-cholinergic agonist oxotremorine (0.0025-0.04 mg/kg). Additional experiments were carried to further characterize the response of rats trained with nicotine (0.1 mg/kg). These animals failed to generalize to compounds from a range of pharmacological classes (i.e., apomorphine, cocaine, chlordiazepoxide, picrotoxin, and quipazine), but there was partial generalization to amphetamine. Mecamylamine (0.5 mg/kg) but not hexamethonium (5.0 mg/kg) blocked the discrimination of nicotine and the generalization to cytisine. Anabasine (1.0-4.0 mg/kg) did not block the response to nicotine. The results support the view that the nicotine cue is mediated mainly through central cholinergic mechanisms. The dose of nicotine used for training has a very significant influence on the characteristics of the cue and 0.1 mg/kg of nicotine may be more suitable than 0.4 mg/kg as a training dose in future work.