This study investigated the potential of thermal remote sensing for estimating ecosystem surface CO2 flux. Ecosystem surface CO2 flux was measured by an eddy covariance method for more than 3 years, in conjunction with thermal and optical remote sensing measurements as well as micrometeorological, soil and plant measurements. The soil was Andisol (Hydric Hapludands), a humic volcanic ash soil, which is the major cultivated soil for upland crops in Japan. The soil surface CO2 flux under bare soil conditions was best correlated with the remotely sensed surface temperature, while air temperature was less well correlated and soil temperature and soil water content were poorly correlated. The relationship was well expressed by an exponential Q 10 function (r 2=0.66, RMSE=0.098). The value of Q 10 and the threshold temperature at which the CO2 flux approached zero were estimated to be 1.47 and 10.0°C, respectively. Results suggested that the soil surface temperature had the dominant effect on the microbial respiration as well as on the physical processes determining the CO2 gas transfer at the soil–atmosphere interface. Remotely sensed surface temperature will provide useful information for investigation of CO2 transfer processes near the soil surface, as well as for quantitative assessment of ecosystem surface CO2 flux.