The author proposes that the implementation of CAT programs in the translation class in degrees other than translation requires a previous study of both the students' actual problems, one of the problems analyzed in this study being finding out the students' most common linguistic difficulties, and the program's potential to tackle or, at least, minimize them. The author presents the results of a case study designed to provide one possible way of assessing the students' difficulties complemented with an assessment of the validity of a popular translation memory system (Translator's Workbench) with a view to testing its possible uses.