Patient and staff perceptions of patient anxiety and depression were determined in 53 dyads of hospitalized cancer patients and their nursing staff. Patient anxiety and depression were assessed with the Hospital Anxiety and Depression Scale. In addition, anxiety was measured by a numerical (0-10) scale. A staff person who had been caring for a certain patient the previous 3 days assessed that patient's anxiety and depression with staff versions of the same instruments. For one subsample (n = 18), staff were also asked to estimate their own hypothetical anxiety were they to have the same disease as the patient. Results indicate that staff overestimated patient anxiety systematically and showed limited ability to adequately assess patient anxiety and depression in terms of rank. Patients and staff agreed about individual patients' levels of anxiety and depression only to a limited degree. Discrepancies between patient and staff ratings of patient anxiety predominantly concerned patients reporting low anxiety levels. Several explanations for these findings are discussed. One possible explanation, the "requirement of mourning" hypothesis, is supported by our findings that staff estimation of patient anxiety was strongly associated with their estimation of their own hypothetical anxiety if they were to be in the patient's situation.