ABSTRACT Correlations of scores on the Modern Language Aptitude Test and the Pimsleur Language Aptitude Battery with FL course marks provided evidence that the tests are best suited to predicting the potential initial rate of FL learning. Attrition, grouping, and self-selection were among factors which reduced predictive power in advanced FL courses by narrowing the range of variability in aptitude and course marks. The testing of three hypotheses revealed that (1) the French students achieved higher aptitude scores and course marks than the Spanish students, (2) achievement of aptitude test scores was not influenced by prior FL study, and (3) the girls were superior to the boys in aptitude and achievement at all levels. The improvement of long-term prediction depends on refinement of objectives, increased precision in defining and measuring FL proficiency, and a continuing reassessment of conceptual relationships. The improvement of instruction should be supported by broader diagnostic use of aptitude measurement as a basis for course differentiation and remedial training. Where circumstances permit, consideration should be given to grouping by sex, especially at the first level.