A large-scale statistical study of the predictors of language gain during study abroad is reported. Subjects were 658 American college and graduate students in the Soviet Union between spring 1984 and spring 1990. Variables examined included student characteristics (age, gender, citizenship, country of birth, place and levels of formal education, highest degree taken, major, prior Russian and other second language training and experience, program type, overseas host institution) and a variety of language measures (proficiency tests, program qualifying exams, learning style or aptitude data). Statistical procedures used are outlined, and results are charted and discussed. It was found that certain student characteristics were predictive of language gains abroad, including gender, experience in learning other foreign languages, and command of grammar and reading skills. Implications are seen for student preparation and selection, in-country program design, and second-language instruction in general. A brief bibliography and author notes are included. (MSE)