Various theoretical perspectives have contributed to the different types of definitions of learning disabilities (LD) that have been developed since the category was first established. To date, there has been little consensus on the definition of learning disabilities. In the present study, three kinds of definitions, ability-achievement discrepancy, low achievement, and scatter, were examined to determine the extent to which there was common variance. The subject were 48 school-identified LD children and 96 non-LD children. Both samples had previously been administered a battery of psychoeducational tests. These data were used to classify each child as LD or non-LD according to each of 14 operational definitions. A factor analysis resulted in two distinct groupings of LD students; low achievement accounted for over four times as much variance as ability-achievement discrepancy. Scatter did not contribute independently to classification. The implications for LD-identification practices are discussed.