Abstract: Fengyun 3 series are second generation polar-orbiting meteorological satellites in China. The first satellite of Fengyun 3 series, FY-3A, is a research and development satellite and is launched successfully at 1100 BT 27 May 2008. The Medium Resolution Spectral Imager (MERSI) is a main payload of FY-3A spacecraft and since June 2008, it has been acquiring daily global data in 20 spectral bands from the visible to the thermal infrared-15 with 1 km and 5 channels with 250 m spatial resolution at nadir. In order to satisfy the requirements of quantitative application, MERSI data need to be geolocated, which can provide accurate latitude and longitude information for follow-up remote sensing productions. The geolocation algorithm of MERSI data, which is adapted by National Satellite Meteorological Center (NSMC), is introduced in details, as well as the error analysis.MERSI rotates a 45° mirror to get 360° views across the orbit direction. MERSI is a paddle broom electro-optical instrument that uses the forward motion of the satellite to provide the along-track direction of scanning. MERSI simultaneously senses, in each band, 10 rows of 1 km detector pixels and 40 rows of 250 m detector pixels. The MERSI detectors are grouped on four focal planes. Detectors for each band are laid on the focal planes in the along-scan direction. MERSI swath is about 2000 km, and generates about 140 GB data per day.According to the characteristics above, parameter method is used for MERSI remote sensing data geolocation. This approach creates the spatial relationship model between the sensed data and the earth based coordinate system, according to MERSI scanning mode, sensing geometry, satellite/sensor's attitude and position. There are 10 coordinate systems and 9 rotated relationships involved in the model. Since 45° mirror brings image rotation, there is a module removing the rotation in the model.When processing data, the line-of-sight vector from each detector of a band is calculated in the instrument coordinate system first. Then the line-of-sight and satellite position are rotated to the earth centered rotating coordinates. The intersection of the line-of-sight with the WGS-84 ellipsoid is calculated. An iterative search process is used to follow the line-of-sight from the instrument to the intersection of the terrain surface represented by a DEM. This geolocation approach for MERSI has been applied to FY-3A data preprocessing system.Five factors that influence the accuracy of results are analyzed. These factors include satellite position/velocity error, satellite attitude error, satellite-instrument installation error, instrument-inner geometry error and instrument thermal distortion.Compared geolocation results with the true remote sensing image using the land-water mask, it shows that the error alone the orbit direction is about 0.167 km and the error along the scanning direction is about 0.058 km. This geolocation approach for MERSI (250 m) achieves accuracy up to 1 pixel.