Objective To study and establish a multiple linear regression model predicting neonatal weightof non⁃obesity and non⁃diabetic maternal. Methods A total of 86 pregnancy maternal were admitted to this hospitalfrom December 2017 to December 2018 were selected as research objects，and they were measured by variousindicator. Several indicators were studied between two groups divided by different average neonatal weight. Then wecan analyze relationship between serious of indicators and neonatal weight by Pearson correlation analysis. A multiplelinear regression model was established which can predict neonatal weight by clinic several indicators and ultrasonicindicators. Then we compared prediction capacity of the mode with other methods. Results There was statisticallysignificant difference（P < 0.05）that weight gains in pregnancy，gestational weeks，uterine height and maternalabdominal circumference，fetal circumference，biparietal diameter and femur length，which were increased alongwith birth. The correlation coefficients of the above indicator on neonatal weight is 0.288，0.366，0.496，0.380，0.419and 0.522，respectively（P < 0.05）. The optimal linear regression model predicted the neonatal weight（F = 22.7，P < 0.005）was 111.2 X1 + 44.5 X2 + 336.1 X3 4454.7（X1，X2，X3 is respectively fetal circumference，fundal height，fetal femur length），which could predict 54% of the neonatal weight.This model of the coincidence rate was signifi⁃cantly higher than other methods（P < 0.05），with higher accuracy for predicting neonatal weight. Conclusion Theprediction regression model is able to predict neonatal weight and the formula is better predictor than other method.