Over the past two decades I have written several articles about assessment procedures in the field of learning disabilities (LD) (Stanovich, 1986, 1988, 1991, 1993, 1996, 1999a, 2000). Many of those articles were centered around the issue of aptitude-achievement discrepancy as a defining feature of a learning disability. In dealing with this issue again here and talking about its future, I was drawn to the title of this essay. The title advertises my frustration with the field on this issue. The persistence of the discrepancy concept in LD signals that the field is not ready to put itself on a scientific footing and that it will continue to operate on the borders of pseudoscience. It is ironic that my other research area is critical thinking, particularly the cognitive processes that lead to pseudoscientific thinking (Stanovich, 1999b, 2002, 2004). The fixation on discrepancy measurement provides a test case of things that I study in that area: confirmation bias (e.g., Nickerson, 1998) and failure to consider alternative theories (e.g., Johnson-Laird, 1999; Stanovich, 1999b, 2004). Imagine that your current HMO covered only the procedures and diseases recognized by the medical profession in 1950. The thought is ridiculous because in such a rapidly developing field as medicine, no one would expect practice to be frozen at the level of scientific knowledge attained 50 years ago. In a less extreme fashion, though, this is what has happened in LD. The field suffers greatly from its tendency to base practice on concepts and psychometric technologies that have been superseded by subsequent scientific advance. I am referring here to the field's persistence in linking the definition of learning disability to the concept of aptitude-achievement discrepancy and identifying aptitude with intelligence test performance. In the following, I confine my comments to reading disability, the most common type of learning disability, and the one where my expertise is concentrated. In the October 2004 issue of the APA Monitor, a leading figure in LD diagnosis is quoted as saying that "the intelligence test is our stethoscope, like it or not" (Kersting, 2004, p. 54), even though there is no research consensus that LD diagnosis using intelligence as a proxy for aptitude is useful or conceptually justified (Siegel, 1989, 1992). Where would one find support for treating the concept of aptitude-achievement discrepancy as foundational--as it has been in practice in the LD field? One would look for evidence supporting four different propositions: (1) that the pattern of information-processing skills that underlie the reading deficits of low-IQ poor readers is different from the information-processing skills that underlie the reading deficits of high-IQ poor readers; (2) that the neuroanatomical differences that underlie the cognitive deficits of these two groups are different; (3) that low- and high-IQ poor readers require different treatments to remediate their reading problems; and (4) that there is differential etiology in the two groups based on different heritability of the component deficits. Why do I call reliance on discrepancy definitions in the LD field an example of pseudoscientific practice? Because there is no strong research consensus supporting any of these four propositions. THE MISSING EVIDENCE Regarding point #1, the preponderance of evidence indicates that the primary indicators of reading difficulty at the word-recognition level do not differentiate poor readers with discrepancy from readers without. For example, several studies have compared the performance of poor readers with high and low IQs and have found that they display equivalent pseudoword reading deficits. Similarly, the two groups display roughly equal deficits in phonological segmentation skills. On measures of orthographic processing, where reading disabled children are generally less impaired, the groups again display no differences. …