Land Surface Temperature (LST) is an important input parameter for many land surface models. Retrieving LST from remote sensing is the main approach for modelling the radiance balance and energy budget at both regional and global scales. Validation of remotely sensed LST is helpful to evaluate its accuracy and stability. Furthermore, it is meaningful for the retrieval and application of remotely sensed LST. Here, first, theories and methods of LST retrieval were reviewed. Second, four validation methods, including the Temperature-based (T-based), Radiance-based (R-based), cross comparison and Time-series analysis, were reviewed and compared. An in-depth examination was conducted for the T-based method from the aspects including the approaches for acquiring the ground truth value, the target LST products, the uncertainty sources. Finally, some important issues in LST validation were discussed.