For the purpose of automatic detection of dust over ocean from satellites, reflection and emission characteristic of water surface, cloud and dust regions are analyzed based on lots of MTSAT satellite data, which show that the brightness temperature at thermal infrared channel 1 BT11 is less than the brightness temperature at thermal infrared channel 2 BT12 in dust sample regions. The visible reflectance of dust sample is generally greater than water surface but less than clouds, while it is opposite at BT11, which confirms that even the atmospheric water vapor content over ocean is relatively high; BT11 is less than BT12 in dust regions by using Libradtran radiative transfer model package. This can be used as an important feature of dust detection over ocean. So an easy dust automatic detection algorithm over ocean by remote sensing using MTSAT satellite data is developed. The thresholds of this algorithm are derived from reflectance, 3 pixel×3 pixel standard deviation of visible channel, BT11 and difference between BT11 and BT12. Then progress of dust storms over the Bohai Sea and the Yellow Sea from March 1st to 2nd, 2008 and from March 19th to 21st, 2010 was monitored both the daytime and nighttime. Result indicates that the method can extract distribution area of dust over ocean and shows transportation process of dust storms objectively. This is in favor of aerosol optical thickness retrieval to estimate the flux of dust sedimentation over ocean.