Rats were trained to discriminate between fentanyl (0.04 mg/kg) and saline in a two-lever procedure. Using a FR 10 schedule of food reinforcement, drug-induced differences between the number of reinforcers obtained under fentanyl and saline conditions were observed. The effect of eliminating these differences on the outcome of generalization tests was investigated by different manipulations. In one group (N = 10), the FR 10 schedule used during saline sessions was changed to FR 6 during drug sessions. In a second group (N = 12), saline sessions ended after the number of reinforcers obtained was equal to the number obtained during the preceding drug session. A control group (N = 10) was trained using a FR 10 schedule under both conditions. Elimination of differences in reinforcement frequency 1) accelerated the acquisition of the discrimination, 2) diminished response bias, 3) flattened the slope and reduced the ED 5 o value of generalization gradients of fentanyl, morphine and sufentanil and 4) increased the EDs0 value of naloxone in antagonizing 0.04mg/kg fentanyl. It is concluded that the unconditioned effects of 0.04 mg/kg fentanyl on response rate in a FR 10 procedure lead to differences between saline and drug sessions which contribute to the apparent discriminative stimulus properties of fentanyl.