The Central Drug Registry in Finland ascertained 5,920 incident cases of Type 1 (insulin-dependent) diabetes mellitus diagnosed under the age of 15 years, during 1965-1984. The incidence was higher for males 29.2/100,000 (95% confidence intervals 28.2-30.2/100,000) than for females 26.1/100,000 (25.1-27.1/100,000). A non-linear increase in incidence with age was confirmed, with peaks at ages 2, 9 and 14 years in males and at 3, 5-6 and 11 years in females. A significant temporal variation in incidence was found, adjusting for age and sex. During 1965 to 1984 the incidence rose by about 57% or by 2.4% annually. However, a non-linear curve with two incidence peaks in 1978 and 1983 would better describe the temporal pattern than a linear trend. There was no significant difference in the temporal variation between males and females. The changes in diabetes risk appeared to affect proportionally all age groups under 15 years. Two possible mechanisms were explored: a calendar period effect vs a birth cohort effect. The calendar time period effect was significant alone and also when adjusted for the birth cohort effect. One the contrary, the birth cohort effect was not significant, when adjusted for the calendar period effect. In conclusion, over the past two decades, the incidence of childhood Type 1 diabetes in Finland has increased by about 57%. The pattern of change was a steady rising background incidence superimposed by sudden outbreaks suggesting environmental causative factors.