The use of test results as a criterion for evaluating the design of an intensive English language program is examined in one Canadian university program in Indonesia. The study was investigating whether: (1) realistic estimates of training needs were being made; (2) data from other tests would enable better prediction of student outcomes, and therefore of student admission criteria; and (3) some guidelines could be established for balancing test preparation and post-course language use in classroom instruction. Subjects were 129 students in 8 academic English classes. Results from one administration each of the Test of English as a Foreign Language (TOEFL) and the Canadian Test of English for Scholars and Trainees were used to predict gain in scores on a second TOEFL administration. Results show the entry TOEFL scores to be misleading as to the homogeneity of the group and to provide an inadequate baseline for assessing student progress. Implications are found for both the design of the training program and for the use of gain scores in program evaluation. Several Canadian initiatives in English second language training and testing are also described. (MSE)