Type I (insulin-dependent) diabetes and genetic heterogeneity Dear Sir, We are glad that Wolf et al. [1] have confirmed several of our previous findings regarding the genetics of Type 1 (insulin-dependent) diabetes [2]. We add some important qualifications and additional data pertinent to their paper.The main issue revolves around the question: is the genetic susceptibility to Type i diabetes due to a homogeneous state (one gene) or a heterogeneous state (several genes or polygenic) and, therefore, several aetiological pathways? An additional question is: is the genetic heterogeneity, if it exists, in the HLA region of chromosome 6 or outside it or both? We have always been a proponent of heterogeneity of both types (within and outside the HLA region), but we think it is important to qualify the strength of the available evidence. As Nell and Spielman [3] have stated, only evidence clearly incompatible with one gene models would represent definite evidence for heterogeneity.Like Wolf et al., we have also found a large excess of the heterozygotes, DR3/DR4 versus other heterozygotes or homozygotes, DR3/DR3 or DR4/DR4 [2]. However, we also found that there was a small excess of homozygotes (DR3/DR3 or DR4/DR4) and the difference between these 'high risk' homo-and heterozygotes was not statistically significant. Since a large number of normal families has never been typed for DR, adequate control frequencies for homozygotes are lacking. Thus, although the apparent excess of heterozygotes DR3/DR4 over homozygotes is suggestive of the existence of two susceptibility genes in the HLA area, it is not proof. A similar conclusion has been reached by others [4].Additional important evidence for genetic heterogeneity in Type 1 diabetes comes from linkage studies. A much tighter linkage between the Type i diabetic trait and HLA in DR3/DR4 probands than in other probands, recently reported by us [5,6] and others [7], constitutes evidence for genetic heterogeneity that may extend beyond the HLA region, incorporating additional non-HLA-linked Type 1 diabetic determinants. This putative non-HLA-linked genetic susceptibility may be related to other associations found between the genetic markers, Gm [8] and KJdd [9], and Type 1 diabetes.The inability of Wolf et al. to find a relationship between HLA and age at diagnosis confirms our findings [2]. Similarly, many groups, including our own [10], have shown that the frequency of HLA recombinants in Type I diabetes is normal. Further, it is true that HLA typing does identify siblings of probands at considerably higher risk of diabetes. However, this risk is still relatively small in absolute value (approximately 10%). For informative studies of the aetiology of Type 1 diabetes, which may lead to rational, preventative strategies, a large number (several thousand) of non-diabetic siblings of diabetic subjects wkh appropriate ages need to be studied prospectively for immune, viral and metabolic parameters. HLA studies are likely to have only a secondary role in these prospective tria...