A model of working and reference memory in rats is described, based on a discrete-trial operant procedure with concurrent spatial matching and nonspatial discrimination components. Working memory was assessed by delivery of food to rats for pressing one of two retractable levers after a delay if that lever had been presented in the prior sample phase of the trial. Reference memory was assessed on other trials by delivering food by pressing the lever illuminated by a cue light after the delay interval. The model was tested with scopolamine (0.10 to 0.56 mg/kg, ip), which reduced matching accuracy in a dose-related manner. Linear slope and intercept estimates of retention gradients showed that intercepts declined and slopes remained unchanged with increasing scopolamine dose, suggesting that cholinergic blockade disrupts encoding processes while sparing retention. In contrast, scopolamine had no effect on nonspatial discrimination accuracy, suggesting insensitivity of reference memory to cholinergic blockade. To compare the effects of scopolamine on spatial and nonspatial discriminations, a second group of rats was trained to discriminate between the spatial locations of two levers.