The validity of a screening process for the early identification of gifted and educable mentally retarded children was examined. The process was originally developed by Lambert and Bower, for the identification of learning handicaps. Twenty-five kindergarten and first grade classrooms were studied. The information collected included the teacher, peer, and self ratings in the screening process, a measure of aptitude, and student characteristics--sex, socioeconomic status, ethnic group, and bilingual status. Objectives were to: study the feasibility of using the assessment materials for initial screening of gifted and of educable mentally retarded pupils; extend teachers' understanding of the emotional and academic needs of their students; and compare the effectiveness of the assessment process with existing information which might be used for screening purposes. Results indicated that: (1) the screening materials tended to "overpredict" or include too many children as candidates for program placement when compared to IQ scores; (2) teachers tended to "underpredict" or exclude too many children as candidates for program placement when compared to IQ scores; and (3) the screening materials appeared to be less vulnerable to socioeconomic status and ethnic or bilingual background than IQ scores. (Author/GDC)