A widely variable overlap ranging from 10 to 92% has been reported in the literature between attention deficit disorder with hyperactivity (ADDH) and learning disability (LD), most likely a result of inconsistencies in the criteria used to define LD in different studies. The following study seeks to more accurately determine rates of LD in clinically referred children. Using a psychometrically reliable methodological approach, it was expected that the rate of LD in ADDH children would be far more modest than previously reported. Subjects were referred children with ADDH (N = 60), children with academic problems (N = 30), and normal controls (N = 36) of both sexes with available psychological and achievement testing. Using a liberal definition of LD, significant differences were found between the groups (ADDH = 38% versus academic problems = 43% versus normals = 8%; p = 0.002). In contrast, more modest rates were found using two more stringent methods of assessment (23 and 17%; 10 and 3%; 2 and 0%, respectively; p = 0.02). Arithmetic-based LD appears to be equally identified by both stringent methods, whereas the liberal definition overidentified children in all three groups. These findings show that a liberal definition of LD overidentifies LD not only in ADDH children but also in normal children.