Pigeons were trained in a 3-choice assay to discriminate among injections of 5.6 mg/kg U-50,488H, 5.6 mg/kg morphine, and vehicle solution. In dose-response tests, subjects rarely responded on the U-50,488H-appropriate key when morphine was administered or on the morphine-appropriate key when they received U-50,488H. A high dose of naltrexone (1.0 mg/kg) completely blocked the morphine cue but failed to block completely the U-50,488H cue. In generalization tests, d-amphetamine primarily engendered saline-appropriate responding. Ethylketazocine produced mixed results, in that moderate doses produced responding on both the morphine- and U-50,488H-appropriate keys, but 3.2 mg/kg engendered primarily morphine-appropriate responding. These results demonstrate the feasibility, but not necessarily the value, of 3-choice discrimination procedures involving mu and kappa agonists and vehicle.