A longitudinal study was conducted in three Maryland counties to determine the predictive validity of the Maryland Systematic Teacher Observation Instrument (MSTOI) and other initial assessment instruments as screening devices to identify kindergarten children with potential reading and/or learning difficulties. The MSTOI was also examined to determine its longitudinal validity as a predictor of children with reading and/or learning problems. Information collected in each county included data on initial tests, follow-up tests, concurrent validity, predictive validity, item analysis, and a discussion of procedures and results. It was determined that the predictive validity of MSTOI compared favorably with other initial assessment instruments. Predictive validity data on MSTOI was developed and its relationship to student progress and retention was established. Recommendations for further study are suggested. Summary tables for individual counties and combined data, descriptions of assessment instruments, and technical notes are appended. (MH)