Abstract: To eliminate interference and improve the transmission rate of cognitive radio multiple-input multiple-output (CR-MIMO) systems, an interference alignment algorithm based on game theory is proposed. The algorithm uses water-filling algorithm for maximum the transmission rate of primary user. Meanwhile, the pre-coding matrix of secondary users is designed for the secondary user signal to fall into free sub-channel of the primary user. Multiple interference links are constituted into a game group to achieve interference alignment of secondary users. Moreover, the power allocation of secondary users is formulated as selection problem of cuckoo's nests, the optimal power allocation is obtained according to the fitness function. Numerical simulation results show that this algorithm can eliminate the interference between the primary user and secondary users and the interference among secondary users. Compared with the maximize-signal-to-interference-plus-noise-ratio algorithm (Max-SINR), the interference alignment algorithm proposed can improve the transmission rate of secondary users about 2 b·s^-1·Hz^-2. Moreover, the transmission rate can also be improved by using cuckoo search algorithm for power distribution compared with the result presented elsewhere.