The Flynn effect, a secular rise in IQ seen throughout the world,was examined on the WISC-R and WISC-III subtests in a longitudinalsample of more than 2,500 school children who were tested between1974 and 2002. Multivariate analysis of variance and multipleregression analyses revealed that all the subtests experiencedsignificant decreases in scores on the introduction of the WISC-III,as expected because of the Flynn effect, with the exceptionof Information and Digit Span.(Mazes was not included in theanalyses because of a limited sample size.) On Picture Arrangement and Coding, however, children who were repeatedly tested onthe WISC-III also experienced significant decreases comparedwith children who were repeatedly tested on the WISC-R. Thesefindings add to the growing literature comparing the magnitudeof the Flynn effect on crystallized versus fluid measures. Implicationsfor special education testing and the current WISC-IV are discussed.