A total of 132 volunteer 10th and 11th grade students participated in an experiment to investigate two variables of computer-based adaptive instructional strategies for concept learning. The first variable tested the hypothesis that selection of number of examples according to on-task information is more efficient than selection according to pre-task information or pre-task plus on-task information. Data analysis showed that the on-task information condition needed significantly less instructional time and fewer instructional examples than either of the other two conditions. The second variable contrasted response-sensitive strategy with a response-insensitive strategy to determine the presentation order of examples within rational sets. Results showed that students in the response-sensitive group not only performed better but also needed less on-task learning time and fewer examples than the response-insensitive group. (Author/CMV)