The purpose of this experiment was to determine whether a faded knowledge of results (KR) frequency during observation of a model's performance enhanced error detection capabilities. During the observation phase, participants observed a model performing a timing task and received KR about the model's performance on each trial or on one of two trials. Delayed retention and transfer tests were used to assess the observer's ability to detect error in the model's performance and in the participant's performance while physically practicing the task. Results indicated a beneficial effect of a reduced KR frequency for performance stability and the ability to detect errors in both the model and the participant's own performance. The results suggest that aspects of the processing mechanism(s) developed in observational learning and related to KR are probably similar to those developed through physical practice. (Contains 3 figures and 3 notes.)