The monthly attack rate with scarlet fever in Japanese children for 24 years, starting with August 1950, showed a characteristic seasonal pattern when the rates were transformed to logarithms. It was well described by a three-term Fourier curve. The average peak of the scarlet fever epidemic occurred on December 3.50 ±13.50 and its minimum on August 14.25±7.81, the two differing by 3.16 fold. The total attack rate for each year varied in apparently random order, no coefficient in a polynomial quadratic equation gave any indication of decreasing or increasing trend within this period. The seasonal incidence pattern found in tetanus during 25 years from January 1950 to December 1974 was fitted with a simple sine curve with a maximum on July 29.19 and a minimum a half year later on January 27.82, the two differing by 1.69 fold. In the case of diphteria, the pattern during 18 years from August 1950 to July 1968 was described by a two-term Fourier curve wih a maximum on November 25.58±5.01 and a minimum on July 25.34±4.83, the two differing by 2.95 fold. The annual totals in tetanus and diphteria incidences decreased progressively over these 25 and 18 year periods, respectively. The linear and quadratic trends between years were significant in both tetanus and diphteria and the data were nonstationary time series. But the seasonal periodicities were changed negligibly by adjustment for the trend between years. The seasonal pattern of monthly attack rate with diphteria vanished during the recent 6 years from August 1968 to July 1974.