Objective: To establish the statistical significance of observed variations over the last decade in the incidence of insulin-dependent diabetes mellitus (IDDM) in the 0- to 19-yr-old age-group and to determine whether incidence has increased in Canterbury, New Zealand. Research Design and Methods: The Canterbury, New Zealand, Diabetes Registry has recorded all incidence cases of diabetes mellitus prospectively since 1982. All IDDM subjects aged 0–19 yr at diagnosis and using insulin are included in the study. Ascertainment is believed to be 100%. Prevalence was recorded at 1 January 1982 and 1 January 1990. Annual incidence for 1982–1990 was determined using age and sex cross-sectional census population denominators. The statistical significance of temporal, age, sex, and seasonal variations in incidence rates was ascertained by Poisson regression models (GLIM statistical software). Results: Prevalence on 1 January 1990 was 115/100,000. Incidence rates during the 9 yr were periodic, with two major peaks—one in the early 1980s, the other in 1989 continuing into 1990. The temporal variation (P < 0.02) was not age or sex specific. Incidence rates for boys were three- to fourfold higher during peak versus trough years, with a peak level of 20.7/100,000 in 1990. For girls, there was less variation, with a peak rate of 21.6/100,000 in 1990. There has been no significant increase in IDDM incidence over time. The mean rate of incidence across all age-groups for 1982–1990 was 12.7/100,000 person-yr. A significant seasonal association to the onset of IDDM was found only in boys, with incidence rates being significantly higher in winter than in summer (P < 0.01). Conclusions: IDDM in Canterbury, New Zealand, presents in cycles of incidence peaks and troughs, each spanning 2–3 yr.