AbstractIn pigeons trained to discriminate 5.6 mg/kg of morphine from saline, cyclazocine, l‐N‐allyl‐normetazocine (l‐NANM, l‐SKF10,047), and ketamine, but not U50,488, produced partial generalization, i.e., a maximum level of drug‐appropriate responding between the levels produced by saline and by the training drug. The generalization gradient of cyclazocine and of l‐NANM, but not that of ketamine, was less steep than the gradient of morphine. Cyclazocine and l‐NANM, but not U50,488 and ketamine, antagonized partially the discriminative stimulus (DS) effects of morphine. Naltrexone antagonized the DS effects of morphine, cyclazocine, and l‐NANM, but not ketamine. Increasing the training dose of morphine shifted the morphine gradient to the right, increased the antagonist effects of cyclazocine and of l‐NANM, and decreased their agonist effects, but did not alter the effects of ketamine. Decreasing the training dose of morphine shifted the morphine gradient to the left and increased the agonist effects of cyclazocine, but did not alter the effects of l‐NANM and ketamine. The full generalization produced by cyclazocine when the training dose of morphine was lowered could be blocked completely by naltrexone and l‐NANM, but not by ketamine. These results suggest that cyclazocine and l‐NANM, but not ketamine, produced partial generalization because of their low efficacy at the receptor that underlies the DS effects of morphine. However, the results obtained with ketamine suggest that partial generalization may also be produced