This experiment investigated self-modeling among 60 children, enrolled in grades 3 and 4, during cognitive skill learning. Children received training on addition and subtraction of fractions. Subjects in one condition (mastery self-model) were videotaped while successfully solving problems and viewed their tapes. Children in the progress self-modeling group were videotaped while learning fraction operations and subsequently viewed their tapes. These subjects experienced initial difficulties, but eventually were successful. Other children were videotaped, but did not view their tapes. Subjects in the fourth condition received training, but were not videotaped. Children in the mastery and progress self-model conditions demonstrated higher self-efficiency, skill, and training performance, compared with subjects in the other two conditions. The mastery and progress treatments did not differ in their effects on children's perceptions of progress in learning. (Author/TJH)