Lake Taihu is a typical eutrophic lake in China, and water surface temperature(WST) becomes one of the most important environmental factors for algae growing in Lake Taihu. The infrared scanner on board Chinese environment and disaster monitoring satellite HJ-1B has a good capability to monitor periodically WST by remote sensing technique. Generalized single-channel method(GSCM), radiative transfer code method(RTM) and mono-window algorithm(MWA) are applied to retrieve WST from six IRS thermal infrared images passing over Lake Taihu, then the retrieved WST is compared with measured one and TERRA/MODIS temperature product. The results show that the WST retrieved by GSCM is a little higher than measured one while that by RTM and WMA is a bit lower, in which the root mean square error(RMSE) of WST calculated by three algorithms is within 1.001 K. The accuracy of WMA is at the best, RTM is at the second good and the third good for GSCM. The RMSE of MODIS temperature product is 1.507 K. The WST histograms retrieved from IRS thermal infrared data by three algorithms showed a positive deviation and a steep normal distribution, suggesting that WST thematic map can display correctly the spatial variability of temperature in Lake Taihu. This will play a guide role to carry out WST monitoring in inland waters by sensor which has only one thermal infrared channel on board satellite.