In order to quantitatively evaluate the response of spring wheat yield to precipitation and temperature changes at different growth periods in Dingxi City, Gansu Province, based on daily precipitation and temperature data from 1971 to 2017 in the study area, the correlations between precipitation and temperature changes with spring wheat yield were analyzed with the APSIM-model simulated growth period and yield of wheat, and the impact of climate changes on the yield of spring wheat was investigated with the quadratic polynomial regression method. The results showed that in the 8 growth stages, 6 climatic factors significantly related to the yield. These factors were precipitation, daily average temperature, daily maximum temperature at jointing-booting stage and grouting-maturity period. The main effect analysis showed that the direct contribution of these six climatic factors were in the order of the during grouting-maturity period’s daily mean maximum temperature, the precipitation, the daily average temperature, the jointing-booting stage’s daily maximum temperature, the daily average temperature, the precipitation. The results of single factor effect analysis showed that: the precipitation at jointing-booting stage showed a quadratic parabolic upward trend to the yield, and the precipitation at grouting-maturity stage showed a quadratic parabolic downward trend to the yield; the yield of wheat decreased with the increase of daily mean temperature from jointing to booting stage in a quadratic parabola, and the daily average temperature during grouting-maturity period showed a quadratic increase to the yield; the yield of wheat would increase with the increase of daily average maximum temperature from jointing to booting stage, but the temperature would increase continuously, after reaching a certain peak, the yield would decrease with the increase of temperature; the daily mean maximum temperature during grouting-maturity period was negatively correlated with yield, and the direct contribution rate to yield was negative, but the quadratic partial regression coefficient was positive, indicating that it was a superimposed positive effect.