The performance of 566 elementary students enrolled in grades 1-6 from Minnesota, Pennsylvania, and Washington, on direct measures of reading, spelling, and written expression was analyzed to determine the rate of students who would be eligible for special education services by each of four discrepancy criteria. Measures were individually administered by a trained examiner. Testing lasted approximately 25 minutes. Stimulus materials were presented in the following order for each student: (1) three isolated word lists, (2) three oral passages, (3) two dictated spelling lists, and (4) two story starters. Results indicated that while a 2.0 to 3.0 times discrepancy from peers was an appropriate level for determining low performance students in grades 3-6, even a stringent 3.0 times discrepancy would identify too many students in grades 1 and 2. The development of a task with a greater number of simple items is suggested as an alternate solution for identifying an appropriate criterion for students in grades 1 and 2. (Author/PN)