The question of whether students perform linguistically differently when writing under examination conditions than when writing term essays is addressed in this working paper. The study was prompted by the feeling by many students that their linguistic performance deteriorates in examination conditions and that this deterioration makes a difference in their grade. Many lecturers claim that they take no notice of how an examination answer is written and grade only on content. This paper examines whether there is in fact a relationship between linguistic performance and grading. The research attempt described centered on the types of prediction identified by Tadros. It is shown that only some of the Tadros types are used by students in the current sample and that so far they have not proved to be a reliable measure of difference. Contains 7 references. (LB)