As a result of advances in technology, genome searches have been carried out for susceptibility genes for type 1 diabetes in humans and in the NOD mouse. These have shown that, in the NOD mouse, diabetes susceptibility is under the control of at least ten separate chromosomal loci. In the human, in addition to HLA and INS, two new susceptibility genes have been localized, IDDM4 on chromosome 11q and IDDM5 on 6q, demonstrating the polygenic nature of type 1 diabetes and the role of HLA as the major locus. Candidate genes at these loci are the subject of current investigation. Genetic and immunological markers of disease may be of value in screening the general population for individuals at risk of developing type 1 diabetes. The predictive power of different screening strategies should be tested in order to work out the potential value to the general population of preventive therapies that are now undergoing clinical trials in high risk 'pre-diabetics'. Type 2 diabetes is genetically heterogeneous, and, since 1992, two distinct genetic subtypes have been identified. The first is defined by mutations in the GCK gene, which cause up to 60% of cases of MODY. The second, designated MIDD (maternally inherited diabetes and deafness), is defined by mutation in the mitochondrial gene for tRNA(Leu(UUR)). MIDD patients are less obese than is usual for typical type 2 diabetes, may present in early adult life or occasionally in childhood and may have been diagnosed as having autoimmune type 1 diabetes, type 2 diabetes or MODY. Typically, patients with MIDD require insulin earlier than do type 2 diabetics without mitochondrial mutations. Genetically complex diseases, such as diabetes, hypertension, cancer and coronary heart disease, are common in most populations. The approaches to the genetic analysis of diabetes outlined in this review are likely to be useful to the genetic analysis of many of these disorders. Progress in this area will have important implications for public health strategies in the next decade and beyond.