The traditional temperature-humidity index (THI) based on observation stations has been widely used to evaluate regional climate comfort, but it is impossible to obtain the spatiotemporal characteristics of larger-scale regional comfort. This study uses MODIS remote sensing data combined with a geographically weighted regression (GWR) model to improve the classic THI model, and the results are compared with traditional interpolation methods to analyze the spatiotemporal evolution of annual and monthly average climate comfort in China from 2005 to 2018. The GWR model uses the land surface temperature (LST), the normalized difference vegetation index (NDVI), and a digital elevation model (DEM) as independent variables to fit the air temperature and accurately express the surface air temperature. According to the average annual THI change from 2005 to 2018, the cooler-more comfortable area increased to 136.11 km², and the annual average comfort level in China changed from cold to comfortable. The Yunnan Province has the most comfortable months, and the central provinces have more comfortable periods than the southeastern coastal provinces. Except for Xinjiang, Tibet, and parts of Northeast China, the spatial distribution of the annual comfort level in China tends to change from comfortable to cold with increasing latitude. Compared with the traditional interpolation method, the THI model based on remote sensing data can more accurately express the spatial distribution characteristics of climate comfort in areas with sparse stations (e.g., northern Qinghai, western Xinjiang, and Hengduan Mountains), especially in mountainous areas in the southwest. This model can also reduce the influence of terrain, elevation, and other factors on the spatial distribution characteristics of comfort.