A study of institutionalized educable mentally handicapped children (EMR's aged 6-7 to 12-6) was designed to evaluate the use of subtest combinations and learning tasks as predictors of achievement compared to single individual and group tests. Fifty-one subjects received a pretest battery of six individual tests, one group test, three learning tasks, and the criterion achievement measures of three subtests from the Wide Range Achievement Test (WRAT) and the New York Achievement Test (NYAT). After 7 months the learning tasks, WRAT, and NYAT were readministered. Results indicated that overall achievement of young institutionalized EMR's will be predicted with higher validity both by a combination of the Auditory-Vocal Association and Auditory-Vocal Automatic subtests from the Illinois Test of Psycholinguistic Abilities, the Picture Vocabulary subtest from the Pictorial Test of Intelligence, and the Number Facility and Perceptual Speed subtest from the Primary Mental Abilities Test (correlation coefficient=.779), and by a regression equation developed from a multiple regression analysis of selected subtests than by any single individual or group test. The better predictors of achievement for the oldest group were subtests which can be administered to groups; individually administered tests were better predictors at younger ages. (RP)