The purpose of this study was to establish specific guidelines for the initial screening and identification of children with learning disabilities in the state of Washington in the elementary grades. Approximately 6,000 children in the state of Washington were screened for learning disabilities using two methods, teacher identification and standardized group testing measures. Next, all students were referred to a qualified clinical team which ascertained which children did, in fact, qualify as learning disabled under Washington State guidelines. Some of the results indicated that teachers consistently over-identified students as learning disabled at all grade levels and also failed to identify some children who were learning disabled. The clinical team verified the classification of all students as learning disabled that the computer had identified, based solely on cutoff levels on standardized tests as specified by the state guidelines. Based on the multiple regression analysis, teachers at the kindergarten level more effectively identified the learning disabled population than at any other grade level. (MKM)