AN INVESTIGATION WAS MADE TO DETERMINE TO WHAT EXTENT TWO METHODS OF ITEM ANALYSIS - NORM REFERENCED AND CRITERION REFERENCED - YIELD THE SAME RELATIVE EVALUATION OF TEST ITEMS. ITEMS WHICH DISCRIMINATED WELL BETWEEN STUDENTS SCORING HIGH AND LOW ON POST-TESTS WERE STUDIED TO SEE IF THEY ALSO DISCRIMINATED WELL BETWEEN PRETRAINING AND POST-TRAINING GROUPS. TWO SETS OF INDEXES WERE COMPUTED FOR THE ITEMS ON EACH OF TWO ARITHMETIC TESTS (ADDITION AND MULTIPLICATION), WHICH HAD BEEN GIVEN BOTH AS PRETESTS AND POST-TESTS IN AN INDIVIDUAL INSTRUCTION PROGRAM IN A PUBLIC ELEMENTARY SCHOOL. IT WAS FOUND THAT THE METHOD OF ITEM ANALYSIS ATTEMPTED IN THIS STUDY (PRETEST AND POST-TEST METHOD) SEEMS TO PRODUCE RESULTS SUFFICIENTLY DIFFERENT FROM TRADITIONAL METHODS TO WARRANT ITS CONSIDERATION WHEN CRITERION REFERENCED TESTS ARE DESIRED. TRADITIONAL ITEM ANALYSIS PROCEDURES WERE DEEMED APPROPRIATE IN THE SELECTION OF NORM REFERENCED MEASURES. (GD)