BACKGROUND: The euglycemic hyperinsulinemic clamp technique is currently regarded as gold standard for measuring insulin sensitivity, but it requires sophisticated equipment and highly trained personnel. We investigated the reliability of short insulin tolerance test as a simple tesl to measure in vivo insulin sensitivity. METHODS: Short insulin tolerance test(SITT) and euglyeemic hyperinsulinemic clamp test were performed at random order in 14 healthy subjects and 10 abnormal glucose tolerant subjects. The plasma glucose disappearance rate(kitt: %/min) after iv injection of regular insulin(0.1U/kg) was determinecl and compared to insulin sensitivitv indices(M, M/I) of euglycemic hyperinsulinemic clamp test. RESULTS: The mean Kitt value of healthy subjects was 3.50+0.75%/min and that of subjects with abnormal glucose tolerance was 2.56+0.56%/min. Changing sampling time from 15 min to 18~21 min and sampling interval from 3 min to 1.5 min had no influence on Kitt value. Kitt values were reproducible in six subjects, with a CV of 8.8+2.0%. There was a highly significant correlations between the Kitt value derived from SITT and M or M/I derived from euglycemic hyperinsulinemic clamp test. There were no significant adverse effects including hypoglycemic symptom while performing SITT. CONCLUSION: SITT is simple, safe, rapid to perform, and provides reliable index of in vivo insulin sensitivity. It seems particularly suitable for studies involving large series of subjects or including repeated evaluation of insulin sensitivity.