The ecosystem service value (ESV) plays a key role to build a global ecological security pattern in order to reveal the evolution characteristics and future evolution of ESV in southwestern China. The revised ESV coefficient was used to quantitatively analyze the dynamic evolution of ESV in the study area from 2005 to 2015. We used the logistics-CA-Markov model to predict the ESV space evolution in 2025. The results showed that:(1) From 2005 to 2015, the total ESV increased by 20.85 billion RMB in southwest China mainly due to the increase of ESV in Qinghai and Tibet waters, and the growth rate of ESV had absolute advantages in Qinghai and Tibet.(2)In the study area, the distribution patterns of ESV in the northwest and southeast were high and low in the north. ESV of different types and grades were spatially converted, and the conversion from low to high grades was obvious. The profit was greater than the loss, and the ecological environment was obviously improved. (3) The Logistic regression analysis showed that the ROC values of all ecological landscapes were more than 0.87, and the fitting results could meet the forecasting requirement. CA-Markov simulated spatial layout of ESV, and the Kappa coefficient was 0.86, which could accurately reflect the spatial evolution pattern of the overall landscape. (4)The spatial evolution map showed the dynamic changes of ESV in various regions of southwest China from 2015 to 2025, reflecting the advantages and disadvantages of the ecological environment. The general trend of the ecological environment in the region tends to be good, but the ecological negative effect of the increase in construction land can not be ignored. It is necessary to rationalize the layout and build an environment-friendly society. In the future, the ecological planning must be carefully laid out. This study provides a detailed spatial evolution map of ESV. It is significance to the implementation of the effective ecological planning and provides a scientific basis for achieving sustainable development.