In two studies, we set out to examine whether the verbal working memory of high and low test-anxious students differed under performance-evaluative threat. In Study 1, 84 schoolchildren completed a backward digit span task under threat or no-threat conditions. In Study 2, 71 schoolchildren completed a backward digit span task in both threat and no-threat conditions. Results showed that the verbal working memory capacity of highly test-anxious students in Study 1 did not change under low or high threat conditions. In Study 2, the verbal working memory capacity of highly test-anxious students decreased under performance-evaluative threat when this condition was taken first but increased when this condition was taken second. To account for the effects of performance-evaluative threat, it is necessary to consider how increased effortful control may compensate for anxiety-induced reduced efficiency when tasks are not timed.