The effectiveness of a strategy for remembering words from a word list was investigated in subjects who were high in test anxiety versus subjects who were low in test anxiety. Sixty seventh-grade volunteers from two urban junior high schools were assigned to the strategy training group or the control group. The strategy group was instructed to cluster words into their component categories, while the control group was urged to repeat the words sub-vocally. A free recall paradigm was used. Subjects recalled a list of twelve words from three categories (fruit, clothes, or animals) until one perfect recall was achieved. Subjects were then given a five-minute coding task and were asked to recall as many of the previously learned words as possible. Two measurements of test anxiety assessed state anxiety at both acquisition and at retrieval. The results supported the hypothesis that strategy training improved the performance of highly anxious subjects, compared to the control group. (GDC)