In many applications, the active noise control (ANC) method has been utilized to cancel unde-sired noises. However, the most of utilizations are only focused on the single reference signal and single output (SISO) system. The research on the MIMO system is very scarce. This thesis de-scribes the MIMO ANC system and its focus is to design and implement the update algorithm of the MIMO ANC system. In the thesis, the FxLMS algorithm is selected as the adaptation algo-rithm and developed for the MIMO case. Its behavior is analyzed in both the frequency and time domains. On the other side, some problems are involved by applying multiple reference signals and multiple outputs. For example, correlated reference signals, feedback effect and the “ill-conditioned”. They could slow down the convergence speeds of the adaptation algorithm, some of them could lead to instability. Therefore, the solutions are also proposed in the thesis to handle these particular problems. Besides the FxLMS algorithm, in order to improve the convergence speed, the new update algorithm, the normalized FxLMS algorithm is also mentioned in the the-sis. In order to verify the FxLMS algorithm and evaluate the performance, the simulations and real time experiments have been conducted. The results are consistent with theoretical analysis and support the conclusions. The MIMO ANC system could attenuate noises in specific frequency range. The previously mentioned problems are solved effectively. Furthermore, the influences of the dominant parameters have been indicated by simulations and implements.