The aim of this study was to examine whether the use of a response-to-intervention (RTI) model to identify specific learning disability (SLD) over-identifies children and youth with population-relative (normative) weaknesses in general cognitive ability (IQs < 90). We compared the overall score on the Kaufman Brief Intelligence Test-Second Edition (KBIT-2; Kaufman and Kaufman 2004) for a group of students with SLD (n = 30) who had been identified in an RTI model with a group of same-age peers (n = 249) in general education. Statistically significant differences were observed between the SLD and general education groups, with considerably lower mean scores for the SLD group. Effect sizes of the mean differences were large. Approximately three-fourths (73.3%) of the SLD group had overall scores on the KBIT-2 that were below the mean of the normative sample and almost half (43.3%) had IQ scores that are below 90. In sum, results of this study support Reynolds' (2009) assertion that use of the RTI model for the identification of SLD over-identifies children and youth with IQs less than 90, thereby fundamentally altering the conceptualization of SLD from the traditional narrow sense (unexpected underachievement) to the broad sense (expected underachievement). A modified hybrid model is presented as an alternative method of SLD identification that addresses the main shortcomings of other current models.