Many anxiety and depression scales are commonly used, although the assumption that they all measure the same construct may be questioned. Thus, researchers have to pay attention to the nature of the scales they use. The Hospital Anxiety and Depression Scale (HADS) was constructed in 1983 to allow a rapid and separate measure of depression and generalised anxiety. Surprisingly, since its introduction, there has been no comprehensive documentation of its psychometric properties. Therefore, as a contribution to assessing the construct validity of the HADS, we conducted a set of confirmatory factor analyses in a sample of 195 healthy students. None of the formerly proposed models fit our data. We were able to split the original Anxiety subscale into two components that we have labelled 'Anxiety' and 'Restlessness', while the original Depression subscale is slightly modified. The results are discussed from both clinical and theoretical points of view.