A study was conducted to: (1) determine the simple and multiple correlation coefficients between selected educational/personal variables and academic achievement at intermediate grade levels as measured by the Iowa Tests of Basic Skills; (2) determine the multiple linear regression equations for predicting individual student achievement as measured by ITBS subtests; and (3) cross/validate the regression equations determined in this investigation. The general method used was the determination and cross-validation of multiple linear regression equations for predicting achievement of intermediate level children from individual and school based data normally available. Variables used were: (1) Dependent: vocabulary, reading, language, and arithmetic; (2) Independent: achievement level, Intelligence, Sex, Social Mobility, Aid for Dependent Children (ADC), age, race, years in school, and learning rate. Sample population consisted of public school children in grades five through eight. Data analysis procedure was Step-Up Multiple Regression Analysis. The SPSS regression routine was used. The highest correlations with post-and pre-achievement scores were on the same scale and intelligence. There was a low relation between learning rate and achievement. In cross-validation, there were five instances of correlations equal to or higher than those originally obtained. Variables that emerged as significant predictors in multiple regression were: pre-achievement, IQ, Sex, and ADC. (Author/CK)