Abstract: [Objective] In order to determine the reasonable planting density of millet, and provide a theoretical basis for agricultural production, [Method] the research methods was by plot test, variance analysis and nonlinear regression. [Result]It showed that planting density had obvious effect on millet traits and yield , with the increase of planting density, millet plant height was reduced, spike length was shortened , stem diameter became smaller, total weight, ear weight, grain weight per ear were reduced, production was firstly increased and then decreased. The suitable planting density was of 22000 plants per mu to 28000 plants per mu. Through SPSS regression curve analysis, we thought that quadratic curve model y=b₀ b₁x b₂x₂ were the optimal model to describe the relationship between millet planting density and yield. The predicted equation of Datong No. 29 in premature district of spring millet was y=102.751 286.622x -54.48x₂. Using the equation, we concluded that the theoretical optimum density was 26305 plants per mu .This theoretical value was in agreement with the measured values. [Conclusion] The millet prediction model which was made by the experimental results is suitable for early maturing area conventional planting density and yield prediction.. But it need further study that the prediction model"s suitability for all the varieties of Foxtail Millet of different ecological conditions in the country.