Subjects selected on the basis of test anxiety scores made a judgement about each face in a series of slides. These orienting tasks involved either abstract traits (e.g. dependability) or physical features (e.g. weight), and decisions were made relative to either an absolute criterion or self-reference. A subsequent recognition test revealed superior performance by low-anxiety subjects and by subjects who evaluated abstract features during study, but there were no depth by anxiety interactions. The reference point for the decision had little effect overall, but the anxiety deficit was more apparent when the absolute reference point was involved.