Objective To assess the exposure levels of heavy metals,including lead,arsenic,mercury,and cadmium,in the local population in Sichuan and Chongqing,China,to compare and analyze the differences in reference intervals(RIs)obtained from direct and indirect sampling methods,and to explore the interchangeability and limitations of these two sampling methods.Methods RIs were obtained by the direct sampling method and the indirect sampling method.In the direct sample method,the levels of blood arsenic,urinary cadmium,urinary mercury,and blood lead levels of 5562 healthy participants aged 22-50 years in Sichuan and Chongqing,China were measured by atomic absorption spectrometry and inductively coupled plasma-mass spectrometry.Using the human biomonitoring(HBM)data,we established RIs for the population by a nonparametric method.On the other hand,in the indirect sampling method,RIs were established via a nonparametric method based on data from the laboratory information system(LIS)of a local hospital after stratifying healthy individuals using a Gaussian mixture model(GMM).Comparative analysis of the RIs derived from the two sampling methods were then conducted.Results The RI for blood arsenic was 0.11-1.3 μmol/L.The RI for urinary cadmium was 0.51-2.80 μmol/mol creatine for adults aged 22 to under 43 years and 0.66-2.96 μmol/mol creatine for adults aged 43-50 years.The RI for urinary mercury was 0.12-1.10 μmol/mol creatine.The RI for blood lead was 14.00-47.00 pg/L for adults aged 22 to under 41 year,16.00-53.38 pg/L for males aged 41-50 year,and 15.00-51.02 pg/L for females aged 41-50 year.Most of the RIs established by the direct sampling method had a narrower range compared to those established by the indirect sampling method,and the RIs established by both sampling methods were partially biased.Conclusions The RIs for blood arsenic,urine cadmium,urine mercury,and blood lead in healthy individuals aged 22-50 years in Sichuan and Chongqing,China were established using both direct and indirect sampling methods,which contributes to a better understanding of environmental exposure to metals in the general population and provides a reference for metal poisoning.For data from the same lab,the GMM-based indirect sampling method demonstrated relatively consistent performance in establishing RIs compared with the direct sampling method.