As one of the important indicators which describe water visibility and optical property,ocean water transparency has good relationship with water quality. Measurement of ocean water transparency from satellite remotely sensed image can compensate for the deficiencies of discrete data, long time and expensive cost of field measurement.Moreover, it can obtain the large-scale distribution of ocean water transparency. So it can be applied to long term observation of ocean water transparency at large scale. Liaodong Gulf was selected as the study area. Ocean water reflectance of MODIS bands were simulated to reduce the disturbance of atmosphere based on in situ filed measurement spectrum. According to the spectral simulation, characteristic bands of MODIS were indentified for water transparency estimation model. The correlationship was respectively ananlyzed in Liaodong Gulf. Then the estimation model of water transparency using MODIS of the area was developed and the model test was also performed. The result shows that in Liaodong Gulf, the band of 667 nm of MODIS has the best correlation with ocean water transparency.Moreover, the estimation model based on single band performs better than band combine approach such as band ratio. Without the incorporation of the classical bands of blue and green bands for oceancolor remote sensing, the estimation model of two-order polynomial of single band 667 nm has the correlation coefficient  r of 0.973(p<0.01) and the relative mean error of model test is 13％. It shows the stability and good forecast performance of the model. As the model was developed based on the dataset of area of the experiment, it has distinct local character. It needs further study to test whether it can be applied to other areas.