In view of the increasingly prominent use in schools and colleges of performance indicators, inspection, appraisal and other forms of performance feedback, it seems important to know what evidence there is that such feedback will produce any improvement in standards, and under what conditions this may be expected to occur. The plausible view that people who receive feedback on a task perform better than those who do not is found to be not universally supported by the empirical evidence. Theories which try to account for the effects of feedback on performance are limited in scope, too vague to be readily operationalized or to enable predictions to be made, and/or supported by only some of the available evidence. An attempt is made to identify the significant variables which mediate the effects of feedback on performance. Important characteristics of the task include its complexity, the balance between the demands of effort and ability, and the availability of other information or instruction. Significant characteristics of the feedback or the way it is given include its relationship to the individual's goals, whether it is ‘ego’ or ‘task’ focused, whether it is ‘norm‐referenced’ or ‘self‐referenced’, whether it is perceived as ‘informational’ or ‘controlling’, whether it is positive or negative, whether it is immediate or delayed, whether it is general or specific, and the credibility attached to it. Significant characteristics of the people receiving the feedback include their level of involvement in the task, their perception of self‐efficacy in relation to the task, their attributional tendencies, locus of control and achievement orientation, as well as their receptiveness to receiving feedback and the adequacy of their original performance. The limitations of the available evidence are discussed and a tentative summary of conditions under which feedback will have optimal effects on teaching performance is put forward.