The reliability and validity of three short forms of the Wechsler Intelligence Scale for Children III (WISC-III) were compared. Each of the short forms was a tetrad composed of two verbal and two performance subtests. The first tetrad was selected based primarily on practical considerations, particularly its brevity to administer and score. The second tetrad was selected on the basis of psychometric and clinical considerations, and the third reflected a blend of psychometric, clinical, and practical factors. WISC-III standardization data (100 males and 100 females at each year of age between 6 and 16) were used to select the short forms, to determine their psychometric properties, and to develop formulas for computing a child's estimated full-scale IQ. All three short forms had acceptable average reliability coefficients above 0.90, but the validity of the brief, practical form was too low to represent a good alternative to either of the other two tetrads. Overall, the short form that represented a compromise between psychometric, clinical, and practical variables seemed to be the best choice. (Contains 4 tables and 27 references.) (Author/SLD)