Juvenile insulin-dependent diabetes mellitus develops in susceptible children exposed to unknown environmental factors. If the genes responsible for susceptibility could be identified, it should be possible to understand the method of injury to beta cells as well as identify the infectious or other agents involved. For a decade it has been known that one or more of the susceptibility genes must be within the major histocompatibility complex (MHC). Unfortunately, there are at least 20 different genes in the complex and it has not been possible to determine which are actually responsible. Therefore, we undertook to apply a new concept and new technology to the problem. Over several years we have shown that the diabetogenic gene(s) are contained within conserved ancestral haplotypes which can then be used as markers of the DNA which must contain the gene(s), whether present in a patient or an asymptomatic carrier such as a parent. This approach avoids the confusion which has resulted from using DR3 or DR4 which are only sometimes associated with the relevant genes. The new technology involves pulsed field gel electrophoresis which allows examination of large fragments of DNA containing all of the MHC, and makes it possible to identify deletions and duplications which were otherwise undetectable. In the first instance we compared two ancestral haplotypes [1,8,3 (8.1) and 18,F1,3 (18.2)] known to contain the relevant genes, and contrasted the DNA with that of another ancestral haplotype [3,7,2(7.1)] which is known to lack these genes. We have shown that there are three major deletions common to the two carrier haplotypes but absent in the protective haplotypes.(ABSTRACT TRUNCATED AT 250 WORDS)