Research on mapping diabetes-susceptibility genes is dependent on several factors, including the existence of a single major gene for susceptibility, genetic homogeneity, and the existence of appropriate clinical material. The power to detect susceptibility genes is dependent on the risks in relatives and the distance of genetic markers from the susceptibility genes. For insulin-dependent diabetes mellitus (IDDM), the best-fitting risk models are those with a single major locus with residual polygenic factors. The major locus effect is likely represented by genes in the HLA complex, because specific genotypes have been found to affect IDDM risk significantly. Thus, mapping the remaining polygenic IDDM susceptibility factors--each of small effect--is a difficult and long task. For non-insulin-dependent diabetes mellitus (NIDDM), the likely risk models result in few genes with moderate effect. Models of NIDDM have significant residual polygenic variation remaining, reflecting the importance of multiple loci with small effect, environmental effects, or genetic heterogeneity; however, the prospects for mapping genes that provide at least moderate susceptibility for NIDDM now appear promising.