Epidemiologic studies have correlated fasting and postload insulin levels with the risk of coronary heart disease, assuming that insulin levels are reliable markers of insulin resistance. However, this assumption has not been systematically studied. The author measured insulin response to an oral glucose load and quantitated insulin resistance using the euglycemic hyperinsulinemic clamp technique to evaluate the correlation between insulin level and the degree of insulin resistance in individuals with varying degrees of glucose tolerance. Subjects were randomly selected from previous population studies done in 1987-1989 at the Department of Medicine of the University of Kuopio in east Finland. Altogether, 50 subjects with normal glucose tolerance, 28 with impaired glucose tolerance, and 54 with non-insulin-dependent diabetes mellitus were studied. Correlations of insulin resistance (whole-body glucose uptake in clamp studies) with fasting or postload insulin levels were remarkably consistent, ranging from -0.58 to -0.74 (p < 0.01) in subjects with normoglycemia. In contrast, corresponding correlations were substantially weaker in subjects with impaired glucose tolerance and non-insulin-dependent diabetes. Among these subjects, only the fasting insulin level correlated significantly with insulin resistance (-0.47, p < 0.05 and -0.48, p < 0.01, respectively). The authors conclude that in population studies, only the fasting insulin level should be used as a marker of insulin resistance, particularly in subjects with abnormal glucose tolerance.