To evaluate the reliability of cause-of-death diagnoses among prostate cancer patients. Information from death certificates obtained from the Swedish Death Register was compared with systematically reviewed medical records from the population-based Swedish Regional Prostate Cancer Register, South-East Region. In total, 5675 patients were included who had been diagnosed with prostate cancer between 1987 and 1999 and who had died before 1 January 2003. The proportion of prostate cancer cases classified as having died from prostate cancer was 3% higher in the official death certificates than in the reviewed records [0.03, 95% confidence interval (CI) 0.02 to 0.04]. Overall agreement between the official cause of death and the reviewed data was 86% (95% CI 85 to 87%). A higher accuracy was observed among men with localized disease (88%, 95% CI 87 to 89%), aged 60 years or younger at death (96%, 95% CI 93 to 100%), or who had undergone curative treatment (91%, 95% CI 88 to 95%). This study indicates a relatively high reliability of official cause-of-death statistics of prostate cancer patients in Sweden. Mortality data obtained from death certificates may be useful in the evaluation of large-scale prostate cancer intervention programmes, especially among younger patients with localized disease.