Some early childhood variables are examined to evaluate their predictive validity. The selection of children needing early childhood Title I services is complicated by the lack of criteria for defining who is educationally disadvantaged and the special problems of early childhood testing and measurement. The study used re-analysis of longitudinal data on children in Head Start Planned Variation and Follow Through programs. The second approach used meta-analysis to synthesize results of studies that examined relationships between early childhood predictors and later outcomes. The strengths and weaknesses of these approaches complemented each other. Methods of selection and their predictive validity were the main focus of the paper. Another factor to be considered included costs of selection procedure. Special problems exist in assessing young children because tests for this age group are often of lower technical quality. Preschool children often lack the physical, intellectual and emotional prerequisites necessary for systematic assessment. Selection bias may result from the use of tests or variables which have different predictive validity for different groups. The importance of prediction stems from the goal of most ECT-I programs: the prevention of educational problems in later schooling. (DWH)