Abstract: An adaptive modulation algorithm is proposed for multiple-input multiple-output (MIMO) systems to maximize transmission rate satisfying transmit power and bit error rate (BER) requirements with imperfect channel state information (CSI). By Lagrange method, the solution of optimization problem in the real field is derived and then adjusted into discrete bits suitable for M-ary quardrature amplitude modulation. Simulation results show this algorithm can provide not only the more robust BER performance to CSI imperfection, but also higher spectral efficiency than the conventional adaptive modulation algorithms based on perfect CSI assumption.