THE HYPOTHESIS OF THIS EXPERIMENT WAS THAT SELF-DIRECTION WOULD BE MORE EFFECTIVE WITH STUDENTS OF HIGH RATHER THAN LOW ABILITY AND INTEREST PROVIDED THEY HAD HAD COACHED PRACTICE IN SELF-DIRECTED USE OF PROGRAMED MATERIALS. SUBJECTS WERE 34 GRADE 9 ALGEBRA STUDENTS. THEY COMPLETED PRE-EXPERIMENTAL ABILITY TESTS AND INTEREST QUESTIONNAIRES. HALF WERE PAID FOR PRE-TRAINING SESSIONS DURING WHICH THEY WERE EXPOSED TO LINEAR (L) AND SELF-DIRECTED (SD) PROGRAMS ON PERMUTATIONS. KEY FEATURE OF THE SD PROGRAM WAS COMPLETE FREEDOM AS TO EXTENT, SEQUENCE, AND METHOD OF STUDY. THEN THE VARIABLES PROGRAM TYPE AND PRE-TRAINING WERE TESTED WITH PROGRAMS ON SET THEORY. ANALYSIS OF COVARIANCE ON GAIN SCORES AND POST-TEST QUESTIONNAIRES SUPPORTED THE HYPOTHESIS AND SUGGESTED THAT SD TECHNIQUES ARE MORE POPULAR WITH HIGH ABILITY STUDENTS AND ARE MORE EFFECTIVE AFTER PRACTICE AND DISCUSSION OF SELF-DIRECTION. THE SD GROUP SHOWED MARKED SUPERIORITY ON THE PRE-TRAINING TOPIC ITSELF. (LH)