Abstract: The performance of conventional beamforming methods for cooperative system degrades severely in the presence of channel state information(CSI) imperfection, which always occurs in practice. Hence it is of interest to develop beamforming algorithm that is robust against CSI uncertainties. This paper addresses the problem of robust beamforming design that maximize signal-to-noise ratio (SNR) at the destination node for amplify-and-forward(AF) based cooperative systems. This algorithm formulates the problem by adopting worstcase design to grantee robust property. Based on extended S-lemma and Schur complement theorem, this originally intractable problem can be transformed to a quasiconvex problem, which can be solved efficiently. Then the robust beamforming vector can be obtained. Simulation results show that the proposed algorithm provides better outage probability performance compared to the conventional algorithm.