PURPOSE: To examine the construct validity of the Swedish version of Hospital Anxiety and Depression Scale (HAD) in women with breast cancer. METHOD: Acquired data on HAD from 727 women who recently had breast cancer surgery, were aged 20-63 years and worked before diagnosis, and had no previous breast cancer (n = 725) were subjected to confirmatory factor analysis examining the viability of three hypothesized measurement models. RESULTS: The analysis showed adequate fit to the data for both bi-dimensional and three-factorial models of HAD. The single-factorial model, however, was shown to have inferior fit to the data. Substantive correlations were found between anxiety and depression in the bi-dimensional model, and negative affectivity and anxiety in the three-factorial model of HAD. CONCLUSIONS: The findings support the utility of scoring procedure based on the original bi-dimensional model, but add indication of co-occurrence of anxiety and depression in this patient population. The discriminant validity of a third factor of negative affectivity in a three-factorial model, however, remains unclear.