Three experimental studies were carried out to examine the ability of groups of learners of English as a second language to solve problems of deductive reasoning (three term linear syllogisms) in their native and second languages. In the first study involving Canadian francophones studying English, subjects solved problems more effectively in their native than in their second language. Their skill in carrying out the experimental task in English was not a function of their competence in that language. In a second study with Canadian francophone subjects the methodology was refined and multivariate statistical techniques were used to analyse the data. The findings confirmed the general pattern of results in the first study and made it possible to locate an information processing difficulty in the second language at the second stage of the problems. Study three was carried out with Arabic-speaking learners of ESL. Once again a similar pattern of results was found. Data from the three studies suggest that second language proficiency scores may not be a very sound predictor of an individual's ability to be educated or to perform certain other tasks in a second language. (Author/AMH)