In the last five years, genetic markers for a large number of diseases have been localised using linkage analysis of DNA polymorphisms in affected families. The site of the genetic defect or defects leading to Type 2 (non-insulin-dependent) diabetes mellitus, a common illness with a major genetic component, remains unknown. This is due, at least in part, to the lack of large well-defined Type 2 diabetic pedigrees suitable for linkage analysis. There are several features of the disease which make large pedigrees difficult to find. The late age of onset of most probands means that informative older generations are often dead, while there is difficulty in detecting disease in younger generations. The diagnostic criteria for diabetes are, as yet, dependent on an arbitrary cut-off along a continuum of plasma glucose. The high prevalence of the disease may also produce problems as, in any given family, diabetogenic genes may be contributed by more than one parent. Varieties of the disease with a well-defined inheritance, such as maturity onset diabetes of youth, are more suitable for linkage analysis but might be due to defects at a different gene locus. Despite these difficulties, once large well-defined pedigrees have been found, linkage analysis using both candidate genes and random highly polymorphic markers is the strategy most likely to find genetic markers for the disease.