Objective To calculate the lower limit of 95% confidence interval of benchmark dose （BMDL） of 3-monochloro-1，2-propanediol （3-MCPD） and its esters for renal tubular hyperplasia effect based on benchmark dose （BMD） method.Methods The data of the most sensitive endpoint of renal tubular hyperplasia in the 2-year chronic toxicity and carcinogenicity study of 3-MCPD were analyzed. BMD model and analysis were used by methods of frequentist， frequentist model averaging and Bayesian model averaging， respectively. Smooth dose response curve was obtained by fitting the nine conventional models， and then the optimal fitting model and BMDL₁₀ estimation value were obtained.Results The best data set was selected by comparing the conservative degree of results across different data sets. Under the same data set， the selection of restricted or non-restricted model parameters， and the selection of single model or model averaging， were key factors to affect the BMDL results of 3-MCPD and its esters. According to model fitting degree and conservative degree， BMDL₁₀ of the most sensitive endpoint of renal tubular hyperplasia of 3-MCPD and its esters was 0.87 mg/kg?BW. The health based guidance value （HBGV） needs to be further determined by the BMDL₁₀ with uncertainty factor.Conclusion The results of this research could be applied for the risk assessment of 3-MCPD and its esters in key foods of China， and further provide scientific advice on the supervision and control of key foods based on the assessment results.