Polymorphisms occur on the average of one out of every 500 base pairs of DNA, and these polymorphisms provide useful markers for genetic analysis. Hundreds of RFLP markers have been mapped at regular intervals throughout the human genome. Diabetes genes have not been mapped with these markers, however, only one MODY family has been partially evaluated. This type of analysis is further complicated if NIDDM is multigenic and/or polygenic. RFLPs have been used to evaluate specific candidate loci for NIDDM, e.g. the insulin, insulin receptor and glucose transporter genes. For these analyses, population and family studies (limited in number) have suggested that none of these loci are major contributors to the genetic susceptibility to NIDDM. In no case, however, could a contribution of 10% or less of these loci be confidently excluded, because of variable penetrance, different degrees of linkage disequilibrium between RFLPs and putative mutations, the frequencies of the RFLPs in non-diabetic populations, and inadequate sample size. The conclusions are clear: either (1) the correct candidate gene(s) has not been found, or (2) sample sizes need to be increased by at least an order of magnitude, or (3) newer methods of analysis must be adopted (e.g. use of extended haplotypes and associations with subphenotypes, or screening with allele specific oligonucleotide probes, denaturing gradient gel electrophoresis or direct genomic sequencing of polymerase chain reaction amplified DNA). © 1991 Bailliére Tindall.