The short insulin tolerance test (SITT) is described as a simple method to measure insulin sensitivity. To investigate the safety and reproducibility of the SITT, 16 healthy volunteers underwent two SITTs within 1 week. Intravenous insulin (0.05 U kg -1 body weight) was injected into an antecubital vein. Blood samples were collected from the contralateral antecubital vein. The insulin-induced glucose disposal rate (K itt ) was calculated from the slope of the regression line of the logarithm of blood glucose against time during the first 3-15 min. Plasma glucose concentrations fell below 2.8 mmol l -1 in 4 of the 32 tests and below 2.2 mmol l -1 in 1 of these 4. Five subjects had mild hypoglycaemic symptoms, three of whom had plasma glucose concentrations below 2.8 mmol l -1 in at least one SITT. The mean K itt was 4.2 % min -1 (range 0.8-8.4) for the first test and 3.4 % min -1 (range 0.1-6.8) for the second test. The mean within-subject coefficient of variation was 30.7 %. We conclude that SITT should be applied with caution especially on insulin sensitive subjects and has poor reproducibility using 0.05 U kg -1 body weight of insulin injection, venous sampling, uncontrolled physical activity and uncontrolled dietary composition. Whether 0.1 U kg -1 body weight of insulin injection and arterialized venous blood sampling as in the original description of this test can improve the reproducibility of the SITT needs further investigation.