One hundred twenty‐three Thai graduate students in Education studying on a one‐year program at The University of Alberta, Canada, completed two administrations of the Test of English as a Foreign Language (TOEFL), after arrival in Canada. TOEFL scores, both total and subtest, together with interview panel ratings obtained in Thailand, were correlated with final grade‐point averages earned in the Alberta program, in an attempt to find the best predictor of academic success. Differing “best” predictors were found in each of the four years of the study. None displayed a high degree of predictive power. The study provided support for the operating of a Summer Language Program for this group of foreign students at a Canadian university. It also supported continued use of objective‐type language tests as a selection device for foreign students, and demonstrated a need for further experimentation with the use of interview panels as a selectio