At present, fire detection based on satellite remote sensing mainly use the mid-infrared band data with the resolution of kilometer level, but the quantitative research on the detection sensitivity of different infrared bands is very few. This is not conducive to give full play to the role of infrared band information in fire monitoring. In this paper, the mixed pixel decomposition method is used for the first time to quantitatively analyze the differences of different infrared bands with resolution of 150 m, 300 m and 1 km in the application of fire spot monitoring. The results show that the sensitivity of the mid-infrared channel with a resolution of 150 m is about 30 times higher than that of the channel with a resolution of 1 km.The far-infrared channel with the resolution of 300 m can detect the fire point of 100 square meters level and the short-wave infrared channel with a resolution of 150 m has obvious response to high intensity open fire area. In this paper, the infrared data of meteorological satellite with a resolution of 1 km and environmental disaster mitigation satellite with a resolution of 150 m as well as 300 m are used to monitor the forest fire in Xunke County of Heilongjiang province in spring and straw burning in Anhui province in summer of 2009, so as to verify the conclusions above. The results show that improving the resolution of infrared band obviously improves the application ability of satellite remote sensing in very small fire spot detection, fire dynamic monitoring, and fire condition assessment.