Abstract: This paper studied the beamforming design in multi-antenna amplify-and-forward two-way relaying system. The beamforming design was formulated as an optimization problem with the aim to minimize the total transmitting power, subject to SNR constraints at the two source nodes. Since the primal problem was nonconex, it was decomposed into three types of subproblems, namely transmit beamforming, receive beamforming and relay beamforming. The transmit/receive beamforming vectors could be obtained via solving the Rayleigh quotient minimization problem. The relay beamforming matrix was obtained by solving the semidefinite programming problem that was formulated by using the semidefinite relaxation method. After solving the above three types of subproblems, an iterative algorithm was proposed to solve the primal beamforming design problem. The convergence of the proposed algorithm was proven by using the monotonic boundary sequence theorem. The computer simulation results show that the proposed algorithm converges to the stable point in a reasonable number of iterations and reduces a great amount of power consumption compared to existing algorithm.