The two-phase study compared two methods of adapting self-instructional materials to individual differences among learners. The methods were compared with each other and with a control condition involving only minimal adaptation. The first adaptation procedure was based on subjects' performances on a learning task in Phase I of the study; the second procedure was based on pre-training abilities. Each procedure was determined by computer decision. The results of the experiment support three conclusions. The first is that training times can be reduced by varying instruction on the basis of learners' abilities. Secondly, the results suggest that a branching strategy can reduce training time further that when both amount learned and training time are of interest, branching is superior to a linear presentation. (Author/LH)