Three pigeons were trained to discriminate among 5 mg/kg pentobarbital, 2 mg/kg amphetamine, a combination of these two drugs at these doses, and saline using a four-choice procedure (amphetamine-pentobarbital group). Three other pigeons were trained to discriminate among 5 mg/kg morphine, 2 mg/kg methamphetamine, a combination of these two drugs at these doses, and saline (methamphetamine-morphine group). After 10 to 13 months of training, the pigeons averaged more than 90% of their responses on the appropriate key during training sessions. In subsequent testing, dose-response curves were determined for the individual drugs, for a wide range of dose combinations of the training drugs, and for two drugs to which the pigeons had not been exposed previously (pseudoephedrine and nicotine). After low test doses of the training drugs, pigeons responded on the saline key. As the dose increased, responding on the key associated with that drug during training sessions increased. When training drugs were combined at doses that were not discriminable when given alone, responding occurred on the saline key. When a discriminable dose of one training drug was combined with a nondiscriminable dose of the other training drug, responding occurred on the key associated with the discriminable dose. When both drugs were given at discriminable doses, responding almost always occurred on the drug-combination key. The response-rate decreasing effects of pentobarbital and amphetamine were mutually antagonized when the drugs were combined, but the rate-decreasing effects of morphine and methamphetamine were not. After low doses of pseudoephedrine and nicotine, pigeons in both groups responded on the saline key. After higher doses of pseudoephedrine and nicotine, responding in the amphetamine-pentobarbital group occurred primarily on the amphetamine key. In the methamphetamine-morphine group, higher doses of pseudoephedrine and especially nicotine engendered more responding on the combination key than had occurred in the other group. The four-choice procedure can reveal subtle effects in the discrimination of individual drugs and drug combinations that are not apparent with procedures offering fewer response alternatives.