ULTIMATE OBJECTIVE OF THIS STUDY IS A SYSTEM IN WHICH INSTRUCTORS MAY IDENTIFY RELEVANT AUDIENCE CHARACTERISTICS AND THEN DESIGN A PROGRAM TO OPTIMIZE LEARNING. MULTIPLE REGRESSION, FACTOR ANALYSIS, AND CROSS-VALIDATING PROCEDURES WERE USED TO ASSESS THE IMMEDIATE AND DELAYED LEARNING EFFECTS OF PROGRAMMED INSTRUCTION. INDEPENDENT VARIABLES STUDIED WERE IN THE AREAS OF VERBAL LEARNING, READABILITY, INCIDENTAL LEARNING, AND RESPONSE DIFFICULTY. 144 PAID STUDENT VOLUNTEERS, PARTICIPANTS IN A PRIOR STUDY, WERE RANDOMLY ASSIGNED TO 3 GROUPS--EXPERIMENTAL, CONTROL, AND CLOZE SCORE. THE 2200 FRAME LINEAR PROGRAM USED TAUGHT CLASSICAL AND OPERANT CONDITIONING. 183 CONSTRUCTED RESPONSE ITEMS FORMED THE CRITERION TEST, AND 6 GAIN SCORES WERE COMPUTED FOR EACH ITEM. COMPLETE STATISTICAL ANALYSIS WAS PREFORMED AND METHODOLOGICAL PROBLEMS WERE DISCUSSED FOR 37 INDEPENDENT VARIABLES. 9 FACTORS IDENTIFIED AND DISCUSSED WERE--QUANTITY OF INSTRUCTION (STEP SIZE), SENTENCE COMPLEXITY, CONCRETE VERSUS ABSTRACT INSTRUCTION, REVIEW, CLOZE SCORE, KEY TEAM, WORD COMPLEXITY, LARGEST FRAME, AND UNNAMED. NEW VARIABLES PROPOSED FOR FUTURE RESEARCH INCLUDE--CATEGORIES OF LEARNING, ESTIMATE OF THE EXTENT TO WHICH RE-CONDITIONS FOR MORE ADVANCED LEARNING ARE SATISFIED, AND INSTRUCTIONAL VARIABLES, SUCH AS FRAME ORDER. (LH)