Objective To establish the reference range of serum human serum insulin levels in our laboratory and investigate measurement stability, Methods The automated Architect i2000 chemiluminescence immunoassay, calibrator and quality control materials were used to measure concentration of serum insulin. Totally 413 cases of healthy adults were enrolled. The physiologic characteristics and the influence of laboratory markers on insulin levels were analyzed. Then the reference range in our laboratory was established. The thermal stability of insulin at different temperatures was investigated. Results Within-run imprecision and between-day imprecision of the automated Architect i2000 insulin assay were 1.67% and 2. 6%, which indicated that this system had good analytical performance. The subjects were classified into 4 groups according to age range ( 10 years). There was no significant difference among four age groups (the median is 5. 6, 5.2, 5.3 and 5. 7 mU/L) (X<'2> = 1. 929,P >0.05). However, the insulin levels were higher in female ( median = 5.7 mU/L) than male ( median = 5.0 mU/L) ( Z = 3.696, P < 0.01 ). The reference range was established based on the 2. 5 and 97.5 percentile values. The reference range of insulin for female was 2. 6-11.8 mU/L and 2. 3-11.6 mU/L for men, which differed from that of other insulin assay. Insulin had a positive correlation with BMI (r =0. 115 ,P =0. 019) and triglyeeride (r =0.143 ,P = 0. 004), and a negative correlation with high density lipoprotein (r = -0.179, P = 0.000). Insulin was stable at 25 ℃ for at least 4 hours or 4 ℃ for 24 hours or could be extended to at least 7 days when stored at -20 ℃. Conclusion Gender is an important physiologic characteristic which has an impact on human serum insuhn level. Insulin reference range should be established according to gender. Insulin immunoreactivity was not very stable at different temperatures.