A study of prediction in three schools, each representing a different socio-economic level, was completed. In each school three predictors, the Lee-Clark Reading Readiness Test, the Bender Gestalt Test (Bender), and the Peabody Picture Vocabulary Test were administered to first grade children in September and used to predict performance on the California Reading Test (Reading Test) and Teachers' Ratings administered during April and May. The Lee-Clark was the best single predictor in all three schools, correlations ranging from .51 to .64 with the Reading Test and .42 to .66 with Teachers' Ratings. The Bender and Peabody did not predict a significant amount of variance in the criteria not already accounted for by the Lee-Clark. Prediction of adequate or inadequate performance for individual students using a cutting score on the Lee-Clark and adequate-inadequate distinctions on the two criteria showed the Lee-Clark to predict failure correctly 73% to 89% of the time. Prediction was improved in the lowest class school by lowering the Lee-Clark cutting score, indicating the possible need for different cutting scores for children of different socio-economic backgrounds.