HLA genotypes were determined and prospective investigations performed on members of 166 families, each including a Type 1 diabetic proband. A computer-based system (Genobase) was created for data storage and analysis. Newly-diagnosed cases ascertained in East Berkshire (incidence rate: 17.8 x 10^-5/year) and additional data from two other studies contribute to some analyses. Major findings include non-random assortment of HLA-B8 within families and a tendency for HLA-identical siblings of diabetic probands to be younger than expected. Discrete multivariate analysis and other methods show no major genetic heterogeneity within Type 1 diabetes. However, there is minor heterogeneity affecting the interaction of age and seasonal variation in incidence of the diabetes. Furthermore, the HLA genotype B8,15 is here shown to be significantly more frequent in Type 1 diabetic probands than expected assuming Hardy-Weinberg equilibrium. With support from other internal and published data, it is concluded that susceptibility to Type 1 diabetes is genetically multifactorial. Increased prevalence of organ-specific autoimmunity is confirmed in the families, and differences in distribution of islet-cell antibodies (ICA) and thyroid/gastric antibodies demonstrated. Five subjects, ICA-positive from the first test, developed diabetes after up to 30 months, demonstrating prolonged latency. Quantitative risk of diabetes is estimated (separately) in terms of ICA results and of HLA genotypes. Non-diabetic siblings of affected probands are investigated for evidence that an HLA-linked immune response gene might contribute to susceptibility. Response patterns to previous polio vaccination give some support, but the excess of thyroid/gastric antibodies in the siblings is not due to such a gene. The dissertation concludes with discussion of aetiological hypotheses and indications for further work.