Evidence of the predictive validity of composite measures of the 'Attitude/Motivation Test Battery' is provided. The composite measures of Motivation, Integrativeness and Attitude Towards the Learning Situation were correlated with three criterion measures, 'Behavioural Intention' to pursue French study, final French grade, and a global measure of French achievement. The data were collected in six different areas of Canada over a two-year period. The composite measure of Motivation was found to be a better predictor of all three criteria, whereas no differences were found between the two attitudinal composites in the prediction of the Behavioural Intention measure and French Grades. Integrativeness was found to be superior to Attitude Towards the Learning Situation for the measure of French achievement. The findings support Gardner's socio-educational model of second language acquisition which proposes that Motivation is a direct determinant of language acquisition and should thus be the most relevant variable of the three.The purpose of this article is to examine the predictive validity of the 'Attitude/Motivation Test Battery' (Gardner, Clement, Smythe & Smythe, 1979). Of specific interest are three composite measures derived from that battery. These are Motivation, Integrativeness and Attitudes Toward the Learning Situation. The relationship of these indices to three criterion measures were investigated. One criterion, Behavioural Intention to pursue French Study, is not a measure of language achievement per se, but it is seen as an important criterion because in contexts where the second language is not readily available such intention reflects a willingness to attempt to develop proficiency in the only situation possible, the classroom. The other two criteria were final French grades which provide an index of success in at least the academic aspects of the language, and an objective global measure of French achievement.The topic of the validity of affective variables involved in second language acquisition is not a new one, and there are many ways of defining validity.