According to federal regulations, children with reading difficulties are eligible for special education services under the learning disability category if they display reading skills that are significantly lower than their scores on intelligence (IQ) tests. Children who are poor readers but do not display this discrepancy are not eligible for special education. A meta-analysis involving 46 studies addressing the validity of this classification of poor readers revealed substantial overlap between the IQ-discrepant and IQ-consistent poor readers. Aggregated effect sizes were in the negligible range for the Behavior (-.05) and Achievement (-.12) domains but in the small range for the Cognitive Ability domain (.30). The latter effects were heterogeneous, with larger estimates showing higher performance by the IQ-discrepant poor readers. The size of the effects could be largely explained by the selection criteria used to form groups, indicating that variation in group definitions across studies accounted for variability in effect size estimates. These results provide little evidence supporting the validity of the IQ-discrepancy classification fundamental to public policy concerning students with learning disabilities and cast doubt on the need for IQ tests in identifying these students.