Faraday rotation (FR) is one of the important error sources for passive microwave remote sensing from space. In this paper, the principle of FR variation is studied. The influence of FR on accuracy for microwave radiometer measurement is analyzed. We concentrate on FR correction both at 1.4 GHz for the orthogonal channel brightness temperature Tv and at 10.7 GHz for the correlative channel brightness temperature U. By using the simulated observational data of spaceborne microwave radiometry at one point in Hainan province in 2006, we compare the effects of two approaches: correction by auxiliary data (IRI model correction) and correction by polarimetric mode (Yueh and Ribó methods). Noise generated by the Monte Carlo mode is included in the simulation. Then a new method of using TEC data released by international GNSS service (IGS) is proposed. For correction of Tv at 1.4 GHz, correction made by polarimetric mode is better than that by auxiliary data. Yueh method is best in effectiveness while IRI model method is worst. For the correction of U at 10.7 GHz, the correction by polarimetric mode is invalid, only correction by auxiliary data is valid. IGS data method greatly improves the correction accuracy and can replace the method of IRI model for nearly real time correction or final data correction.