BACKGROUND: Chronological changes in mortality and case fatality rates from Kawasaki disease covering an extended period in Japan are still unknown. METHODS: We analyzed 679 deaths of patients in Japan whose underlying cause was Kawasaki disease, by using the data of vital statistics between 1972 and 1998. RESULTS: The male-to-female ratio of the number of deaths was 2.07 and the mean age at death in males was higher. Two unusual increases in the epidemic years, 1982 and 1986, were observed in the chronological changes of the number of deaths. The mortality rate of males was higher than that of females, with a few exceptional years, and the annual mortality rates were high in three epidemic years. The age-specific mortality rate was highest in infants under one year of age. Prefectures with high mortality rate clustered in some regions. The case fatality rate decreased annually, declining to as low as 0.2% among those who were born in 1986 and thereafter; and unusual increases in the case fatality rate affected by these three epidemic years were not repeated. CONCLUSIONS: The case fatality rate from Kawasaki disease in Japan decreased during the 27 years of observation: improvements in treatment might account for this. J Epidemiol 2003;13:246-250.