The use of Temperature, humidity, sunlight, precipitation, wind and other meteorological elements were selected to establish an comfort index for tourism climate, and 19812010 climate data and the RCP4.5 scenario predictive data of Hainan Island were used to analyze the tourism climate index for 1981-2010, 2011-2040, and 2041-2070. The results showed that the index changed significantly in different months, and with temperature and precipitation increasing, the comfort level for tourists will be improved in January of 2011-2040 and 2041-2070 as compared with the 1981-2010 average level. The comfort level is reduced in April, July, and October. The indices for tourists attraction and climate were used to establish a climate-tourism flows model based on 5 counties of large tourist flows (Haikou, Sanya, Wanning, Danzhou, Wuzhishan) in Hainan Island. Assuming the tourism attraction index does not change, the model was used to analyze how climate change affects the tourist flows by RCP4. 5-estimated data, The results showed that with the decrease of the tourism climate index, the quantity of tourists spending a night locally would decrease in Haikou and Sanya.