In parallel experiments designed to find the genetic determinants of type 2 diabetes in Oji-Cree, we identified several linked chromosomal regions, using genomic scanning, in addition to a private diabetes-associated mutation, namely HNF1A G319S, using candidate gene sequencing. The genome scan did not identify the region harboring HNF1A as being linked with diabetes. Also, the HNF1A mutation, when used directly in sib-pair linkage analysis, was not linked with diabetes. However, HNF1A G319S was very strongly associated with diabetes, predicted the clinical severity of diabetes, and performed well as a diagnostic predictive test for diabetes in the Oji-Cree. Despite the failure of linkage analysis to identify HNF1A as a determinant of type 2 diabetes, we feel justified in interpreting that G319S has a very important pathogenic role in Oji-Cree diabetes, based upon the highly suggestive association studies. The probable etiologic heterogeneity of type 2 diabetes in the Oji-Cree created a situation in which association analysis was much more sensitive to detect a relationship between HNF1A S319 and diabetes than was linkage analysis. The effectiveness of linkage analysis will probably be limited in study samples that have an even greater complexity of genetic background and/or disease etiology. Thus, the absence of linkage does not always mean that a genomic variant is not an important determinant of a complex disease. Furthermore, our experience confirms the value of using several complementary strategies to identify susceptibility genes for a complex disease.