The correlations between numbers of tsutsugamushi disease patients and meteorological elements were analyzed for 11 years from 1982 to 1992 in Gifu Prefecture, Japan by using regression analysis. The number of patients in early winter was closely correlated independently to both the mean of the minimum temperatures from 11th May to 31st July and the mean of the maximum temperatures in November in the same year. Regression coefficients (R2) were 0.689 and 0.560, respectively. On this basis, an equation for prediction of the number of patients in early winter was designed as follows: N = [e (j - 17.6) + 2.3(v - 13)] x j x v/156 (prediction formula 1) N:predicted number of patients in early winter j:the mean of the minimum temperature from 11th May to 31st July v: the mean of the maximum temperature in November e:the base of the natural logarithm (= 2.718...) The number of patients in early winter was also closely correlated to j in an equation of the fifth degree (R2 = 0.930). N = 22.524656384 j5 - 2218.23705 j4 + 87272.992 j3 - 1714734.329 j2 + 16825634.235 j - 65963810.254 Based on these formulas, the temperature in early summer has a significant effect upon the prevalence of tsutsugamushi disease in early winter.