Family history data were collected in 1203 consecutive admissions of insulin-dependent diabetes mellitus cases to the Children's Hospital of Pittsburgh between Dec. 31, 1964 and Jan. 1, 1981. This report deals with two issues: the closeness in time of dates of onset of diabetes in multiple sibling case families and the similarity of ages of onset within these families. The age-specific incidence rates among siblings were 6-18 times higher than in the general population. Contrary to other reports, this study does not find an increased risk to siblings during the first year or two after the onset of the index diabetic case. The mean duration between cases is 6.1 years. Fitting a log-normal distribution to the periods between cases verified the long median incubation period (4.35 years) with a high degree of variability (dispersion factor = 3.38). However, the log-normal was not a good fit to the data (p less than 0.005). There is a significant correlation for age of onset for pairs of affected siblings within families (0.25). However, it is shown that the similarity in age of onset within families may be a function of the greater similarity to ages of siblings within a family. It is noted that secondary cases are more often younger children in the family, particularly so if the index child is of school age at the time of onset.