Recent studies have consistently revealed reduced feedback frequency facilitates motor learning, especially in variable condition. But the reasons behind reduced feedback frequency effect still remain to be determined. The main purpose of this study was to examine if error detection, that is important for the recall of movement knowledge, can be enhanced by reduced knowledge of results (KR) frequency. Two experiments were used to test this hypothesis. Experiment 1 randomly assigned participants (n=28) to one of the following conditions: 100% KR or 50% KR. Using a hand dynamometer, participants practiced 3 targets of force production. The three targets were determined by calculating 30, 50, and 70 percent of the participants’ maximum grip strength. The results revealed reduced KR frequency enhanced learning in retention but not performance in acquisition, which replicated the finding of our previous research. More interestingly, the participants in 50% KR group increased accuracy in estimating their performance compared to 100% group in retention, F (1, 26)=4.72, p<.05. It strongly suggested that 50% KR frequency enhanced error detection. In Experiment 2, participants (n=30) practiced three goals of angle displacement (45, 60, and 75 degrees). In particular, participants from both 50% KR group and 100% KR group were asked to estimate their results every other three trials during acquisition. Results showed no differences in performance and learning between the two KR conditions. This indicated the manipulation of performance estimation in acquisition improved the error detection of 100% KR group.Keyword(s): performance, research