The language training program for Peace Corps participants is investigated. The objectives were: to investigate the validity of the Modern Language Aptitude Test (MLAT) for predicting end-of-training Foreign Service Institute (FSI) language proficiency; to discover the relative difficulty of various target languages; to develop tentative expectancy tables to enable prediction of a trainee's likely proficiency at end of training; and, to outline the need for and nature of an ongoing program of Peace Corps language research. Target language, total instructional hours, MLAT scores, initial and final FSI proficiency ratings were the variables considered. Findings indicated that MLAT has modest, though statistically reliable, predictive power for all languages; greater validity for common than for exotic languages; greater predictive power for shorter instructional periods (less than 280 hours) if language is disregarded; and taking language type and training time into consideration, greatest predictive power in common language projects involving 280 or more hours of training. Systematic proficiency differences were observed among languages, implying differences in relative difficulty for American students. More vigorous and innovative research is recommended to identify factors of importance in language training. Some thirty languages are grouped and listed in order of difficulty, and expectancy tables indicate the likelihood of attaining various levels of end-of-training proficiency. (Author/LR)