The correlation coefficient between an ability test and achievement test is necessary in order to estimate the effects of the regression of IQ on achievement when calculating a discrepancy between them for the purpose of classifying a child as learning disabled. Weighted mean correlations and their respective variances were computed across studies for one of 11 ability-achievement test pairs using a meta-analysis procedure called Validity Generalization. The results indicated that there is no "global correlation coefficient" that will accurately represent the relationship between all ability and achievement tests. Furthermore, critical data necessary to adjust correlations and their respective variances for statistical artifacts were not reported in the literature. Thus, the results obtained from the subgroups could be due to capitalization on chance.