Abstract: Objective To establish the mathematical models of stature estimation for Sichuan Han female with measurement of lumbar vertebrae by X-ray to provide essential data for forensic anthropology research. Methods The samples, 206 Sichuan Han females, were divided into three groups including group A, B and C according to the ages. Group A （206 samples） consisted of all ages, group B （116 samples） were 20-45 years old and 90 samples over 45 years old were group C. All the samples were examined lumbar vertebrae through CR technology, including the parameters of five centrums （L₁-L₅） as anterior border, posterior border and central heights （x₁-x₁₅）, total central height of lumbar spine （x₁₆）, and the real height of every sample. The linear regression analysis was produced using the parameters to establish the mathematical models of stature estimation. Sixty-two trained subjects were tested to verify the accuracy of the mathematical models. Results The established mathematical models by hypothesis test of linear regression equation model were statistically significant （P<0.05）. The standard errors of the equation were 2.982-5.004 cm, while correlation coefficients were 0.370-0.779 and multiple correlation coefficients were 0.533-0.834. The return tests of the highest correlation coefficient and multiple correlation coefficient of each group showed that the highest accuracy of the multiple regression equation, y=100.33+1.489 x₃-0.548 x₆+ 0.772 x₉+0.058 x₁₂+0.645 x₁₅, in group A were 80.6% （±1SE） and 100% （±2SE）. Conclusion The established mathematical models in this study could be applied for the stature estimation for Sichuan Han females.

Key words: forensic anthropology, forensic radiology, tomography, X-ray computed, body height, lumbar vertebrae, regression analysis