The Corrective Feedback Paradigm (CFP), a computer-based instruction model, addresses the problem of inefficient sequencing of items in a set to be learned by suggesting the use of an increasing ratio review schedule for presenting drill items. With this system, items answered correctly are either removed from the list entirely or replaced at the end of the stack of items; if an incorrect answer is given, the missed item is inserted in the stack of drill items according to a specified spaced review schedule. To determine the effectiveness of increasing ratio review for teaching generalizable concepts, 51 subfreshman, enrolled in one of three Algebra I classes at a university laboratory high school, utilized the PLATO system to study the generalizations suggested by the relationship between the equations for curves. For purposes of the experiment, three drill paradigms were used: (1) increasing ratio review with generalized item generation (IRR-GEN); (2) increasing ratio review with fixed items (IRR-FIX); and (3) immediate review only with generalized item generation (IMM-GEN). Pre- and posttests were given to subjects who were randomly assigned to three treatment groups. It was found that the systematic review of difficult items helped the IRR-GEN subjects to remember the concepts over time better than the immediate review treatment helped the IMM-GEN group. There was also a significant difference in posttest performance between the IRR-FIX and IMM-GEN groups. These differences support the claim that increasing ratio review techniques can be successful in enhancing traditional drills to teach generalizations. PLATO system directions and the study posttest are appended, and a list of references is provided. (JB)