As an important climate resource, near surface air temperature plays an essential role in driving the vegetation growth, monitoring agro-meteorological disasters and global change. The spatially continuous air temperature images, which can be better derived from remotely sensed data than interpolated of meteorological data from limited ground stations, are required as the important input parameters by several environmental models at the regional scale. To serve and support the scientific research and business work concerning the distribution of air temperature at regional scale, in this paper, we firstly introduced the main five categories of methods to estimate air temperature from remotely sensed thermal data, including simple statistical approaches, the advanced statistical approaches, the temperaturevegetation index (TVX), the method based on surface energy balance and the one based on temperature lapse rate. Then, the approaches of maximum, minimum and mean air temperatures estimation were discussed in more details, followed by the comparison of air temperature estimation at different temporal and spatial scales. Finally, the problems and prospects for remotely sensed estimation of air temperature were summarized.