NIDDM is usually characterized by β-cell failure and decreased insulin sensitivity. It has been reported that a high proportion of African-American NIDDM subjects are insulin sensitive. To examine this issue, we deter mined insulin sensitivity (S(I)) in 479 NIDDM subjects by minimal model analyses of frequently sampled intra venous glucose tolerance (FSIGT) from the Insulin Resistance Atherosclerosis Study (IRAS), a large multicenter study of insulin sensitivity and cardiovascular risk in African-Americans, Hispanics, and non- Hispanic whites. The African-Americans and non-Hispanic whites were sampled in Los Angeles and Oakland, California. The non-Hispanic whites and Hispanics were sampled in San Antonio, Texas, and San Luis Valley, Colorado. We defined the proportion of insulin-sensitive (S(I)) subjects as ≤1.61 min-1 · pU-1 · ml-1, which is above the median for nondiabetic subjects of all ethnic groups in the IRAS. Using this definition, the proportion of insulin- sensitive diabetic subjects was very low in all ethnic groups (non-Hispanic whites [14.3%] vs. African- Americans [6.5% ], P = 0.039 in Los Angeles and Oakland; non-Hispanic whites [6.8%] vs. Hispanics [4.9%], P = 0.737 in San Luis Valley and San Antonio). These results were also similar in newly diagnosed mildly hyperglycemic diabetic subjects. In addition, these results were not affected by the adjustment for differences in obesity, body fat distribution, and severity of hyperglycemia. Even in nonobese subjects (with BMI <30 kg/m2), the proportion of insulin-sensitive subjects (S(I) ≤1.61 min-1 · μU-1 · ml-1) was low (3.6-9.7%). The acute insulin response (AIR) was significantly higher in African-Americans than in non-Hispanic whites; there were no ethnic differences in AIR between Hispanics and non- Hispanic whites. There were no significant ethnic differences for non- insulin-mediated glucose disposal (S(G)). We conclude that the number of insulin-sensitive NIDDM subjects is low and similar among non-Hispanic whites, Hispanics, and African-Americans in the U.S.