The accuracy of the satellite-measured reflectance is a key question for data processing and oceanographic applications.A hyperspectral satellite remote sensing reflectance evaluation model(HRSREM) was developed to evaluate the accuracy of the satellite measured reflectance.The model can compute the total reflectance at the top of the atmosphere(TOA) according to observation conditions of satellites,based on a radiative transfer model with the consideration of multiple scattering effects and atmospheric absorption effects.The performance of the HRSREM model was examined by Gordon’s algorithms,showing that the relative errors of the Rayleigh scattering reflectance and the aerosol scattering reflectance are less than 2%.The model can also compute the sky reflectance which can be validated by in-situ measurements.The two sky reflectances match well with a spectral average error of 5.4%.The relative error of the total reflectance of the model,verified by sea-viewing wide field-of-view sensor(SeaWiFS) data,is about 3.5%.Therefore,the total reflectances at TOA,computed by the model,can be taken as reference values to evaluate the accuracy of satellite reflectances.The model was used to evaluate the accuracy of the hypersptral satellite(Hyperion) remote sensing data.The Hyperion reflectance matches the total reflectance of HRSREM very well at visible and near-infrared bands with an average error of 7.3%,while the status of calibration coefficients at shortwave infrared bands are not stable with a large spectral average error of 63.5%.The reflectance evaluation of a moderate resolution imaging spectrometer(CMODIS) data indicated that relative errors are large,especially at near-infrared bands with relative errors more than 100%.The calibration coefficients of CMODIS,obtained from laboratory measurements,are not reliable.The CMODIS data should be recalibrated for oceanographic applications.The performance of the HRSREM model is effective in evaluating satellite data and its algorithms can be easily modified in order to evaluate the accuracy of other ocean color satellite sensors. The accuracy of the satellite-measured reflectance is a key question for data processing and oceanographic applications. A hyperspectral satellite remote sensing reflectance evaluation model (HRSREM) was developed to evaluate the accuracy of the satellite measured reflectance. The model can compute the total reflectance at the top of the atmosphere (TOA) according to observation conditions of satellites, based on a radiative transfer model with the consideration of multiple scattering effects and atmospheric absorption effects. the performance of the HRSREM model was examined by Gordon’s algorithms, showing that the relative errors of the Rayleigh scattering reflectance and the aerosol scattering reflectance are less than 2%. The model can also compute the sky reflectance which can be validated by in-situ measurements. The two sky reflectances match well with a spectral average error of 5.4%. relative error of the total reflectance of the model, verified by sea-viewing wide field-of-view sens or (SeaWiFS) data, is about 3.5%. Wherefore, the total reflectance at TOA, computed by the model, can be taken as reference values ​​to evaluate the accuracy of satellite reflectances. The model was used to evaluate the accuracy of the hypersptral satellite (Hyperion) remote sensing data. The total reflectance of HRSREM very well at visible and near-infrared bands with an average error of 7.3%, while the status of calibration coefficients at shortwave infrared bands are not stable with a large spectral average error of 63.5%. The reflectance evaluation of a moderate resolution imaging spectrometer (CMODIS) data indicated that relative errors are large, especially at near-infrared bands with relative errors more than 100%. The calibration coefficients of CMODIS, obtained from laboratory measurements , are not reliable. The CMODIS data should be recalibrated for oceanographic applications. The performance of the HRSREM model is effective in evaluating satellite data and itsalgorithms can be easily modified in order to evaluate the accuracy of other ocean color satellite sensors.