Background : To examine the problems involved in writing practice of death certificates, we compared the determination of underlying cause of death for vital statistics using recorded underlying cause of death in issued death statistics. Methods : We collected 688 mortality certificates issue in year of 2,000 from 3 university hospitals. And we also collected vital statistics from ministry of statistics. The causes of death were coded by experienced medical record specialists. And causes of death determined at ministry of statistics for national vital statistics were mapped to causes of death recorded at each death certificates. The rate that underlying causes of death for vital statistics were derived from underlying causes of death recorded at issued death certificates were analysed. Results : 64.5% of underlying cause of death for could be derived from underlying cause of death recorded at issued death certificates, 8.6% derived from intermediate cause of death, and 3.9% derived from direct cause of death. In 23% of cases, underlying cause of death could not be derived using issued death certificates. The rate that underlying cause of death for vital statistics could be derived from underlying cause of death recorded at death certificates was different between 3 university hospitals. And the rate was also different between death certificates and postmortem certificates. We classified the causes of death using 21 major categories. The rate was different between diseases or conditions that caused death too. Conclusion : When we examined the correctness of death certificate writing practice using above methods, correctness of writing could not be told as satisfactory. There was difference in correctness of writing between hospitals, between death certificates and postmortem certificates, and between diseases and conditions that caused death. With this results, we suggested some strategy to improve the correctness of death certificate writing practice.