The discriminative functions of the response–reinforcer relation may contribute to the changes in response rates that occur when reinforcement is delayed. These properties were investigated in three experiments with pigeons using a discrete‐trials conditional discrimination procedure. A concurrent variable‐interval schedule was arranged on two side keys during a sample component. The key peck that ended the schedule (the sample response) initiated a delay with either a stimulus present throughout the delay interval (full signal), a stimulus present only during the first second of the interval (partial signal), or no stimulus present (unsignaled delay). The delay was followed by a choice component where one alternative was reinforced if the left sample response produced the choice component and the other if the right sample response produced the choice component. Accuracy was high with a full signal; slightly lower with a partial signal; and lowest without a signal. The results parallel the effects of similar delays programmed in conventional reinforcement schedules. This in turn suggests a possible discriminative effect of the response–reinforcer relation in the control of behavior by (delayed) reinforcement.