Two studies evaluated the differential behavioral effects of instructions and feedback in matching-to-sample procedures. In Experiment 1, 20 college students received true or false instructions and trial-by-trial or delayed feedback in three phases. In a fourth, final phase the instructions remained the same, but feedback changed from trial-by-trial to delayed, or from delayed to trial-by-trial. In Experiment 2, half of another 20 participants received true instructions during three phases, followed by false instruction in a fourth phase; the other half of the participants received false instructions during three phases, followed by true instructions in the fourth phase. Feedback sequences were as in Experiment 1. The results of both experiments revealed historical effects of instructions and feedback. Most participants demonstrated strong instructional control, overriding the control by contingencies. These results suggest that the present procedure offers optimal possibilities to make the differential effects of instructions and feedback on human behavior clearly identifiable when conditional discrimination tasks are used.