Abstract: [Objective]In order to improve the forage sorghum yield level in autumn idle land, the linear regression prediction model of sowing times and biological yields of forage sorghum was statistically constructed. The correlation between sowing times and biological yields of forage sorghum was simulated and compared by conducting field plot experiment and employing linear regression analysis method. [Results]The sowing times had great influence on the biological yields of forage sorghum, and the biological yields were decreased with the delay of sowing times. The regression model between sowing times and biological yields was constructed： the regression model of fresh yields and sowing times is yfresh=196.646-4.625 x, and that of hay yields and sowing times is yhay=58.253-1.423 x. Between July 23rd and August 30^th, the fresh yields and hay yields of forage sorghum were averagely reduced by 4.625 kg/20 m^2 and 1.423 kg/20 m^2 respectively when the sowing times were delayed by every one day. [Conclusion]The regression model constructed in this study provides scientific basis for improving the production level of forage sorghum in autumn idle land.