The purpose of this study was (1) to determine if academic achievement of Anglo-American, Mexican-American, and Negro students in junior high school can be predicted from a set of intellective and nonintellective variables and (2) to ascertain if ethnic group membership is related to variables used for best prediction of academic achievement for those students. McGuire's model of academic behavior served as a theoretical framework for the study. Eleven variables were used to predict ninth grade achievement for 90 Anglo-American, 77 Mexican-American, and 91 Negro students. All subjects attended integrated urban schools located in communities of comparable lower socioeconomic status. Zero-order, part, and multiple correlation procedures were used to analyze the data. The criterion was approached in two manners: grade nine quality point average and grade nine QPA residual (part correlation techniques were used to remove the variance associated with grade eight QPA predictor). Intellective variables included convergent thinking determined by the California Test of Mental Maturity, Short Form, divergent thinking measured by Seeing Problems, symbolic aptitude assessed by Concealed Words, maintaining a convergent set indicated by STEP Listening, and acquired response repertory determined by grade eight QPA. Nonintellective variables were conformity motivation measured by the Survey of Study Habits and Attitudes, self-esteem assessed by the Checklist of Trait Names, internal control determined by the Intellectual Achievement Responsibility Questionnaire, birth order, expectations of significant others indicated by Peer Acceptance: Academic Model, and sex of student. Intellective and nonintellective variables were related significantly to grade nine QPA for each group, but conformity motivation and grade eight QPA were the only variables with significant correlation coefficients for all groups. Grade eight QPA was the best predictor of grade nine QPA for each group. Values of £ for this variable were 0.749 for Anglo-Americans, 0.684 for Mexican-Americans, and 0.809 for Negroes. Significant multiple correlation coefficients with grade nine QPA residual were obtained for all groups. Conformity motivation and intemal control made significant contributions to the multiple JI of 0.456 for Anglo-American pupils. Conformity motivation was the single variable contributing to the £ of 0.274 for Mexican-American students. Divergent thinking was the only significant predictor of grade nine QPA residual for Negroes and the correlation coefficient for this variable was 0.224. Results confirmed academic achievement of Anglo-American, Mexican-American, and Negro students can be predicted in junior high school and ethnic membership tended to influence the choice of predictors for those students. Comparisons among groups showed Mexican-American students to be least predictable and the importance of familial and other culture specific attitudes and values was postulated as a possible cause of the negative and nonsignificant relationships of predictors to QPA for those students. Similarities among correlates of achievement for Anglo-American and Negro students were evident. The importance of educators considering ethnic membership when planning learning experiences for students was discussed.