Although learning disabled children have been traditionally defined as those who show a discrepancy between achievement and potential, previous early identification studies have predicted generalized low achievement rather than discrepancy from potential. The present investigation addressed this definitional issue by comparing the results obtained in an early identification study when a discrepancy criterion was employed with the results obtained when a generalized low-achievement criterion was used. Subjects were all first-grade children from normal classrooms in six elementary schools in rural southern Illinois. They were administered a variety of cognitive, neuropsychological, language, and behavioral measures upon entering the first grade. Follow-up reading and math measures were collected the following spring. Results demonstrated that the use of an unadjusted achievement raw score criterion results in misdiagnosing a significant number of children as learning disabled, as defined by the discrepancy criterion and in failing to identify a significant number of children who are, indeed, learning disabled. Also, somewhat different sets of predictor variables were associated with these two criteria. Comparisons of the findings with past research and implications for further research are discussed.A learning disability is commonly defined as a failure to learn at a level commensurate with potential despite adequate intellectual ability, sensory capacity, psychological adjustment, and environmental opportunity (Benton, 1975;Eisenberg, 1978;Ysseldyke, Algozzine, Shinn, & McGue, 1982). This definition distinguishes between poor academic performance that is "expected" (i.e., on the basis of intellectual ability, sensory capacity, etc.) and poor performance that is "unexpected" (i.e., a learning disability). Epidemiological research has demonstrated the presence of The authors would like to express their gratitude for the cooperation of the following individuals in making this project possible: F. E.
