In this article we discuss research bearing on the traditional use of the IQ-achievement discrepancy to define specific reading disability. We initially review the evidence presented by Rutter and Yule (1975) in support of this practice, and then discuss results from subsequent studies that have questioned the reliability of their findings. We also discuss results from more recent studies demonstrating that the IQ-achievement discrepancy does not reliably distinguish poor from normal readers, whereas language-based measures do reliably distinguish these groups. We highlight results from a study we recently completed, in which it was found that IQ scores did not differentiate between poor readers who were found to be readily remediated and poor readers who were difficult to remediate. In view of the convergent evidence against the use of IQ scores to define specific reading disability, we suggest that the IQ-achievement discrepancy definition of this disorder be discarded.