Adaptive beanfforming algorithms continue to be a research hotspot of smart antennas.Among adaptive beamforming algorithms,non-blind adaptive algorithms attract most researchersinterests because they have better feasibility,stability performance and they are not affected by the accuracy of DOA estimation.The performance of non-blind adaptive algorithms used in TD-SCDMA communication system has not been studied and simulated in-depth yet.This paper tries to simulate and compare three typical non-blind adaptive algorithms including Least Mean Square algorithm (LMS),Recursive Least Squares algorithm (RLS),the Sample Covariance Matrix Inversion algorithm (SMI) for TD-SCDMA system.The simulation results show that all three algorithms can mitigate interferences,save power and improve the quality of sigoal.The simulation also indicates that better performances and applicability are obtained when the weights are estimated using the RLS algorithm than the LMS and SMI algorithms.