It has long been noted that late onset diabetes is common in the parents and siblings of individuals who develop diabetes before the age of 30 [1]. In keeping with this, later studies have shown an apparent overlap between Type I and Type II diabetes in these families [2±7]. Type II diabetes has been observed more often in relatives of individuals with Type I diabetes than in control groups [3,4], and siblings of children with Type I diabetes are more likely to be affected if a parent has Type II diabetes [5,6]. An overlap between Type I and Type II diabetes was also suggested by a study of probands with Type II diabetes in which those with a family history of Type I diabetes showed a phenotype more characteristic of Type I diabetes [7]. These observations indicate a genetic link between Type I and Type II diabetes but the method of phenotyping used in previous studies requires further consideration.Classification of diabetes by treatment [8,9] has now been superseded by classification according to cause [10,11]. As a result, it is now recognised that Diabetologia (2002) Abstract Aims/hypothesis. Previous studies have reported an excess of Type II (non-insulin-dependent) diabetes mellitus in parents of children with Type I (insulindependent) diabetes mellitus. We set out to characterise the clinical and immunogenetic features of diabetes in parents of affected children, and to test the hypothesis that there is no excess of Type II diabetes within this population. Methods. Clinical details were collected from 3164 parents of 1641 children with Type I diabetes participating in the Bart's-Oxford study of childhood diabetes. Islet cell antibodies, antibodies to GAD and IA-2, and HLA class II genotype were determined in a subset of this group. Individuals were assigned a classification of Type I diabetes on the basis of clinical features and measurement of islet autoantibodies. Results. Of 184 parents with diabetes, 138 (75 %) were on insulin. At least one islet autoantibody was detected in 90 (59 %) of 152 parents tested, and of 116 who were HLA-typed, 23 (20 %) had the highest risk genotype HLA-DRB1*03-DQA1*0501-DQB1*0201 / DRB1*04-DQA1*0301-DQB1*0302. Of 46 non-insulin-treated parents, 12 had islet autoantibodies. Of all parents, 141 (4.5 %) were therefore classified as having Type I diabetes, and 31 (0.98 %) as Type II diabetes; 12 could not be classified because of missing data or samples. Conclusion/interpretation. Autoimmune diabetes can present late and without immediate need for insulin treatment in parents of children with the disease. Previous studies have categorised this as Type II diabetes. Our study suggests that there is no excess of non-autoimmune diabetes in the families of children with Type I diabetes. [Diabetologia (2002) 45:495±501]