The purpose of this study was to investigate the effect of learning model's KR relative frequency during observational learning of simple motor task for observer. Forty female university students served as participants and were randomly assigned to one of the following observation group:(a) 100% KR group; (b) 50% KR group; (c) 20% KR group; and (d) NOKR group. Simple arm-outward movements was experimental tasks. Two-way ANOVA last factor repeated and Tukey HSD method were used for statistical analysis. In AE scores, interaction between groups and retentions was found. In ten-minute retention, NOKR group and 20% KR group were better accurate than that in 100%KR group. In 100%KR group, ten-minute retention was less accurate than that in one-minute retention. In NOKR group, ten-minute retention was better accurate than that in one-minute retention. In VE and CE scores, there were not interaction between groups and retentions. Furthermore there were not significant different of both groups and retentions. In | CE | scores, interaction between groups and retentions was found. In ten-minute retention, NOKR group and 20%KR group were less bias than that in 100% KR group, NOKR group was less bias than that in 50%KR group. In 100%KR group, ten-minute retention was higher bias than that in one-minute retention. In NOKR group, ten-minute retention was lower bias than that in one-minute retention. It is concluded that, during observational learning, no KR or lower relative frequency of KR from learning model is a learning variable of simple motor task.