The training dose was progressively reduced (faded) using rats in a drug discrimination task in order to determine whether the discriminative stimulus effects of phencyclidine HCI (PCP) were separable from its response rate effects. Rats were initially trained to discriminate PCP from saline in a two-lever food-reinforced operant discrimination procedure. After initial training at 3.0 mg/kg, the rats were subsequently retrained at 1.5, 0.75 and 0.375 mg/kg. Stimulus generalization testing followed training at each training dose. Although discriminative control was developed by lower doses of PCP during the fading procedures, control by previously effective higher doses was not diminished. The number of training sessions required before reaching criterion levels of performance and the number of errors during training (i.e., incorrect lever selections) increased with decreases in training dose. The results indicated that fading of the training dose of PCP caused marked parallel shifts in the dose-effect curves for PCP's discriminative stimulus effects although its response rate effects were left unaffected. The ED50 for PCP's discriminative stimulus effects decreased with decreases in training dose although the ratio of the ED50 to the training dose remained relatively constant. The ED50 for the response rate effects, however, remained relatively constant with decreases in training dose, and the ratio of the ED50 to the training dose markedly increased. It was concluded that the discriminative stimulus and the response rate effects of PCP are separable and, hence are under control by different determinants.