Water surface temperature is one of the important environmental parameters at the breeding aquatics area in shore. The highly dynamic monitoring of this temperature will help to arrange aquaculture production and breeding production. Remote sensing technology is a very efficacious means for monitoring water surface temperature in high dynamic condition. At present, weather and ocean satellites which have lower spatial resolution are widely used in water surface temperature remote sensing monitoring, with the spatial resolution being 1 000 m. However, water bodies in shore are influenced by land and ocean as well as the spatial changes of water surface temperature, and hence satellites data which have higher spatial resolution are needed to meet the monitoring for water surface temperature in high dynamic condition. HJ-1B/IRS has a 4-band infrared sensor (IRS). One of the bands is the thermal band which can be used to retrieve water surface temperature. In this paper, data of about ten images of HJ-1B/IRS thermal band obtained from 2008 to 2009 and the atmosphere measurement data at the time when the HJ-1B satellite passed through the area were collected. Based on the monowindow algorithm and referring to the temperature product from EOS/MODIS, the authors established a model for water surface temperature retrieval from HJ-1B/IRS data. Moreover, a comparison was made between two retrieval methods for surface temperature from HJ-1B/IRS thermal band and from the temperature product by EOS/MODIS. The results show that the absolutely average temperature error from the monowindow algorithm and from the temperature product by EOS/MODIS is 7.84℃, while that from the method proposed in this paper is 0.83℃. The model for water surface temperature retrieval from HJ-1B/IRS data established in this paper was applied in the area of Liaodong Bay for the purpose of carrying out dynamic monitoring of water surface temperature.