HIGH SCHOOL CHEMISTRY UNIT TESTS AND A PRETEST DESIGNED TO PREDICT STUDENT SUCCESS ON A STATE CHEMISTRY EXAMINATION WERE CONSTRUCTED. HIGH SCHOOL SCIENCE TEACHERS AND UNIVERSITY SCIENCE EDUCATORS USED BLOOM'S TAXONOMY OF EDUCATIONAL OBJECTIVES TO CLASSIFY TEST QUESTIONS ACCORDING TO THE DEGREE THAT THEY INVOLVED RECALL, COMPREHENSION, APPLICATION, OR HIGHER COMPETENCIES. RESULTS FROM TRIAL USE OF THE INSTRUMENTS WERE ANALYZED THROUGH ITEM ANALYSIS. REVISED FORMS OF THE TEST WERE USED AS REGULAR EXAMINATIONS FOR APPROXIMATELY 1,200 STUDENTS DURING 1 SCHOOL YEAR. CORRELATIONS BETWEEN UNIT TESTS, AND UNIT TESTS AND PRETESTS WERE DETERMINED. MULTIVARIATE ANALYSIS WAS USED TO IDENTIFY COMBINATIONS OF TESTS THAT WERE THE BEST PREDICTORS OF STUDENT SUCCESS ON THE STATE EXAMINATION. CORRELATIONS WERE ALSO DETERMINED FOR STUDENTS' TOTAL SCORES ON DIFFERENT CLASSES OF QUESTIONS ON ALL TESTS. INTERCORRELATIONS BETWEEN THE UNIT TESTS AND THE PRETESTS RANGED FROM .45 TO .78. THOSE FOR THE UNIT TESTS RANGED FROM .61 TO .78. MULTIPLE CORRELATION COEFFICIENTS FOR DIFFERENT COMBINATIONS OF PREDICTING INSTRUMENTS RANGED FROM .75 TO .84. INTERCORRELATIONS FOR THE FOUR CATEGORIES OF TEST QUESTIONS RANGED FROM .80 TO .87. THIS ARTICLE IS PUBLISHED IN THE "JOURNAL OF RESEARCH IN SCIENCE TEACHING," VOLUME 4, ISSUE 2, 1966. (AG)