An analysis on cause-specific mortality at the provincial level provides essential information for policy formulation and makes it possible to draw hypotheses regarding various diseases and causes of death. Although the mortality level and causes of death at the provincial level are determined by the multiple effects of socioeconomic, cultural, medical and ecological factors, this study primarily intends to examine similarities and differences of cause-specific mortality at the provincial level. Utilizing the registered death and the registered population as of 1998, the delayed death registration and unreported infant deaths were supplemented at the provincial level and age-standardized death rates and life tables were calculated. Regarding the mortality level due to all causes, major findings were as follow: (1) For both sexes as a whole, Seoul showed the lowest mortality level, and Jeonnam showed the highest mortality level; and (2) The differences of the mortality level among provinces were greater for males than females and for those less than 65 years than those 65 years and over. Regarding the cause-specific mortality level revealed in all indicators (cause-specific age-standardized mortality rates and the probability of dying at birth due to a specific cause for males, females, and both sexes combined respectively), the major findings were as follow: (1) The mortality level due to heart diseases was the highest in Busan and the lowest in Gangweon; (2) The mortality level due to liver diseases was the highest in Chonnam; and (3) The mortality level due to traffic accidents was the highest in Chungnam and the lowest in Inchon. As the mortality differentials at the provincial level are related to various factors, exploratory statistical analysis is attempted for the 25 explanatory variables including socioeconomic variables and 90 mortality variables. Mortality due to all causes are related to socioeconomic variables. Among cause-specific mortality, mortality due to liver diseases and traffic accidents is related to socioeconomic variables. Finally, the need to improve the quality of death certificate is discussed.