The risk of insulin-dependent diabetes mellitus (IDDM) was examined in siblings of an unselected population (n = 194) of newly diagnosed diabetic individuals less than 30 yr old. From 1 July 1984 to 30 June 1987, diabetic subjects (proband) identified within a geographically defined area of southern Wisconsin were studied. IDDM occurred among siblings of probands in 13.5% of families and was associated with proband age at diagnosis. The highest risk was found for diabetic subjects less than 10 yr old at diagnosis (P = 0.04). We did not find an association between sibling IDDM and proband sex, HLA-DR3/4, duration of symptoms, or ketosis at diagnosis. In addition, the odds ratio (OR) for the association of IDDM in the proband with IDDM in parents and second- and third-degree family members was examined by case-control methodology. Diabetic subjects were matched to two types of control subjects (friends and general population) by age stratum and sex. The OR for IDDM was not increased significantly if parental IDDM or non-insulin-dependent diabetes mellitus (NIDDM) was reported. However, there were very few parents with diabetes among diabetic or control subjects. In 6.4% of diabetic subjects, one parent had IDDM, 54% of whom were fathers. In 4.3% of diabetic subjects, one parent had NIDDM, and 57.1% of these were fathers. The OR for IDDM was significantly increased if second- and/or third-degree relatives had IDDM (OR diabetic subjects vs. general population 2.33 [P less than 0.05)] or NIDDM (OR diabetic subjects vs. friends 2.05 [P less than 0.01]).(ABSTRACT TRUNCATED AT 250 WORDS)