The paper describes the application of discriminant analysis to the identification of academically gifted elementary grade students. The principles of discriminant analysis are explained to provide a weighted profile of scores across predictor variables which are then compared with a criterion measure. An illustration of the method's used with 64 possibly gifted students in Garden City, Kansas, is given. The screening procedure focused on three predictor variables: student standardized achievement test scores, teacher ratings, and student performance and response rates on a series of nonverbal tasks. Additionally, 85 experts in gifted education were asked to indicate the prevalence of specific behaviors in gifted and nongifted children and the relative importance of each characteristic. Experts rated cognitive factors as the best predictors of giftedness. Teachers identified four factors as most predictive of giftedness: creativity, diligence, high level cognition, and criticism of self and others. Among nonverbal tasks found to be predictive were the Maze Tracing Speed Test, the Identical Pictures Test, Addition/Subtraction Test, Story Sequence Test, and Number Comparison Test. A score of 125 or higher on the Wechsler Intelligence Scale for Children-Revised was used as the criterion measure to classify students into gifted and nongifted groups. Use of the method to identify students for individualized testing would have led to a 26% reduction in individual testing time while being 90% accurate in the identification of gifted students. (DB)