Knowledge of results (KOR) is most frequently cited as the reinforcer in self-instructional systems. The printed answer in a programmed text, for example, is supposed to reinforce the response the student emits previous to observing that answer. Some other possible reinforcers are briefly discussed in this paper before the literature on KOR in self-instruction was selectively reviewed. The review was organized as a search for evidence that KOR might appropriately be called a reinforcer. Studies comparing programs with and without feedback were examined; the weight of evidence from these global studies was that feedback did not enhance learning, as measured by immediate post-test scores or by retention tests. In at least one case there seemed to be a decrement in performance traceable to the presentation of feedback. Studies in which "schedules of reinforcement" were varied similarly failed to show effects that would be expected if KOR were acting as a reinforcer. One major study involving delay of KOR did report the effect expected when delivery of a reinforcer is delayed. Other studies on delay do not replicate this finding. Finer grained analyses of student behavior and KOR begin to reveal specific conditions under which KOR seems to be acting as a reinforcer. (Author/AMM)