Five low-functioning mentally retarded subjects were taught initially to insert tokens in a slot only when an auditory-visual complex stimulus was presented. The auditory component was a spoken reinforcer name (e.g., "Food"), and the visual component was a flashing red light in the slot opening. Later, the name was presented without the light on probe trials. Three subjects did not respond, suggesting that their token insertions had been controlled by the visual but not by the auditory component of the auditory-visual complex. These subjects then received a fading program designed to establish auditory stimulus control. The program was successful with two subjects. The third subject required post-program remedial teaching, but ultimately demonstrated exclusive control by the spoken name. The results suggest methods for establishing auditory stimulus control with this population and highlight variables that may influence the reliability of such control in some subjects.