OBJECTIVE: Previous studies of depression and anxiety during the year before death have reported different findings. We therefore aimed to study depression and anxiety in patients who had died from cancer and had previously attended cancer clinics. METHODS: We analysed routine data on 4869 deceased patients who had completed the Hospital Anxiety and Depression Scale (HADS) as part of their routine cancer care. The HADS data were linked with demographic, cancer and mortality data from national registries. We used data from all HADS completed in the last year of life to investigate the relationships between mean depression (HADS-D) and anxiety (HADS-A) scores and the percentages of high scores (≥11 on each subscale) and time to death (Analysis 1). This analysis used multivariable linear regression with cubic splines and robust standard errors to allow for multiple HADS from the same patients. We also investigated within-patient changes in scores (Analysis 2) in a subset of patients who had completed more than one HADS. RESULTS: In Analysis 1, modelled mean HADS-D scores increased by around 2.5 and the percentage of high HADS-D scores increased from 13% at six months before death to 30% at one month before death. Changes in HADS-A were smaller and occurred later. In Analysis 2, similar patterns were observed in individual patients' HADS scores. CONCLUSION: Depression should be looked for and treated in patients with cancer and a prognosis of six months or less, in order to maximise the quality of patients' remaining life.