This research note deals with the acquisition of motor skills, specifically with the optimal use of feedback on goal achievement termed knowledge of results, KR for the maximization of learning and retention. A great deal of evidence suggests that enhancing the amount or quality of KR can improve performance in a session where KR is present, but limited evidence suggests that this might not be effective on a delayed criterion test when KR is removed -- a typical goal of many Army training settings. Two variations of KR were studied relative frequency - the proportion of trials receiving KR, and summary KR - where KR is given about an entire set of trials. In each case, alterations in KR which degraded training performance relative to a condition with KR on each trial actually produced enhanced performance on a delayed no-KR retention test. A third paradigm examined the learning of error-detection capabilities as a basis for these effects. Overall, six experiments suggest that enhancing KR in acquisition may generate KR overreliance, preventing the learning of important features of the task which are critical for retention of performance when feedback is removed or degraded e.g. in markmanship, Our results have implications for improved feedback in training.