Variables derived from two measures of classroom behavior, direct observation and teacher rating, were used in regression designs with measured intelligence to predict a measure of academic achievement of 69 kindergarten, first, and second grade children. Results suggest that measures of classroom behavior contribute variance to the prediction of academic achievement measures independent of, and in addition to, measured intelligence. Differences in the predictive classroom behaviors are noted among the various grades. Discussion focuses upon the possible functional relationship of classroom behavior and measured academic achievement and developmental, environmental, and other influences on this relationship. (Author)