Based on Landsat5 TM and Landsat8 OLI/TIRS data, the land surface temperature of Nanjing city is retrieved by radiative transfer equation, and the thermal landscape pattern evolution is analyzed by landscape pattern indices. Then the thermal environment change of Nanjing in 2024 is predicted by CA-Markov model. The results show that, from 2000 to 2017, the area of high temperature zone in Nanjing increased by 456.55 km2 and 149.75 %, and the area of mid-temperature zone, low and sub-low temperature zone decreased continuously. On the patch-class level, the dominance of high temperature patch in thermal landscape pattern was rising rapidly. From 2000 to 2010, the fragmentation and complexity of the high temperature patch decreased, which is contrary to that from 2010 to 2017. On the landscape level, the fragmentation degree of thermal landscape pattern in Nanjing decreased, the area of various patches tended to be uniform, and the degree of aggregation among patches increased. It is shown that the simulation result of thermal environment of Nanjing in 2017 based on CA-Markov model has a high simulation accuracy, with the average error of each heat island type less than 6% and the Kappa index of the model of 78.36%. The prediction results of thermal environment of Nanjing in 2024 show that the area of high temperature and sub-high temperature zone will increase by 94.96 km2 and 126.91 km2 respectively, and the area of mid-temperature zone, low and sub-low temperature zone will continue to decrease.