Identification of children with learning disabilities is based on the notion of a significant discrepancy between ability and achievement. Current federal guidelines do not specify the extent of such discrepancies but indicate they should be "severe". Local education agencies have adopted criteria suggested by professionals or formulated their own operational criteria for identification of learning disabled children. In this research, we examined the extent to which identification as learning disabled is a function of the criteria used to define "severe discrepancy", and the extent to which different classifications would result from use of different criteria. A school district made identification decisions for 51 students referred because they were experiencing academic difficulties; they found 24 students LD. The school identification decisions, based on application of a severe discrepancy on the Woodcock-Johnson Psycho-Educational Battery, did not correlate with decisions based on application of the federal definition. Implications for decision-making practices are discussed.