AN INVESTIGATION OF THE VALIDITY OF INTELLIGENCE AND OTHER TESTS USED IN THE DIAGNOSIS OF RETARDED CHILDREN WAS PERFORMED. EXPERIMENTAL SAMPLES CONSISTED OF 101 CHILDREN SELECTED FROM SPECIAL CLASSES FOR EDUCABLE MENTALLY RETARDED (EMR) WHOSE AGES RANGED FROM 6.9 TO 10 YEARS AND WHOSE IQ SCORES RANGED FROM 50 TO 80. THE TESTS EVALUATED WERE (1) STANFORD-BINET, (2) ILLINOIS TEST OF PSYCHOLINGUISTIC ABILITIES (ITPA), (3) PEABODY PICTURE VOCABULARY TEST (PPVT), (4) PICTORIAL TEST OF INTELLIGENCE (PTI), (5) COLOURED PROGRESSIVE MATRICES (CPM), AND (6) PRIMARY MENTAL ABILITIES TEST (PMAT). THESE TESTS WERE ADMINISTERED TO THE GROUP WITH VARIOUS LEARNING ABILITY TASKS. A STATISTICAL ANALYSIS SHOWED THAT THE PMAT, STANFORD-BINET, PTI, AND THE ITPA WERE VALID, WHILE THE OTHER TWO WERE LESS VALID. THE USE OF A COMBINATION OF ONE GROUP TEST AND ONE INDIVIDUAL TEST WAS ALSO SUPPORTED BY THESE RESULTS, AND THE CPM AND THE PPVT ARE NOT ADEQUATE SUBSTITUTES FOR MORE COMPREHENSIVE INSTRUMENTS THAT PREDICT LEARNING ABILITY AT AN EARLY AGE. (RS)