A study was developed to evaluate the effectiveness of a preinstruction retention index (designed to maximize recall of mathematical concepts by predicting the idiosyncratic number of examples per mathematical concept required by each student). Subjects, 27 female and 26 male eighth grade students, were administered a retention measure through computer-assisted instruction (CAI) and randomly assigned to one of three treatment groups, each providing similar CAI mathematical concepts and different methods of determining the number of examples per concept provided (variable, choice, or fixed). A two-way analysis of variance, with sex and treatment group as the independent variables, was performed; results showed that females in the "variable" group performed significantly better on retention measures than did females in "choice" or "fixed" groups, and that males in the "choice" group performed better on retention measures than did males in the other two groups. In addition, it was found that the use of the preinstruction index resulted in overall better retention for females but not for males, indicating the possible usefulness of such an index in mathematics instruction and a need for further research into the sex variable in retention. (Author/SP)