A delayed conditional discrimination (DCD) task in the rat was modified by requiring a response on an admission lever at the end of each delay. This requirement proved effective in precluding the use of positional cues as mediating behavior. Furthermore, validity of the procedure was assessed by examining how performance changed as a result of: length of the delay, retroactive and proactive interference, and encoding time. Results showed that log d, a measure of stimulus discriminability, decreased on longer delays (Experiment 1); decreased when an interfering stimulus was presented during delays (Experiment 2); decreased when the intertrial interval was made shorter (Experiment 3); and decreased when the sample stimuli were presented for a shorter period of time (Experiment 4). Log b, an index of bias, remained low throughout the study, indicating that no significant response bias was present. Taken together, the results support the notion that this modified DCD task is a valid model for working memory that effectively precludes the use of positional cues as mediating behavior in the rat.