A significant increase in insulin resistance has been implicated in many human diseases and the absence of a simple, accurate, reproducible measurement of in vivo insulin sensitivity has become a major concern. In order to evaluate the reproducibility of insulin sensitivity measured by the modified insulin suppression test, 12 healthy young Chinese men were subjected to the same test two weeks apart. After three days on a standard diet and an overnight fast, somatostatin (350 micrograms/h), insulin (25 mU/m2/min) and glucose (240 mg/m2/min) were infused concomitantly for three hours. A steady state plasma glucose (SSPG) concentration achieved during the last 30 minutes of infusion represented the measurement of insulin sensitivity. Comparisons between the metabolic clearance rate of insulin (MCRi), plasma total triglyceride and lipoprotein cholesterol fractions on two different days were carried out. The results indicated that mean SSPG concentrations on Day 1 (5.73 +/- 0.43 mmol/L) correlated with mean SSPG concentrations on Day 14 (5.51 +/- 0.38 mmol/L; r = 0.82, p < 0.002). The relationship slope did not differ from 1 (0.74, p > 0.05), the intercept was close to the origin (1.24 mmol/L, p > 0.05) and the mean coefficient of intra-individual variation was 10.3%. There was no difference between the MCRi for Day 1 and Day 14 (529 +/- 26 vs 526 +/- 24 mL/m2/min, p = NS), with a mean coefficient of intra-individual variation of 6.9%.(ABSTRACT TRUNCATED AT 250 WORDS)