A new finite state machine (FSM) partitioning approach is proposed in this paper. Genetic algorithm (GA) is employed to search the optimal or near optimal solution. A new cost function is used to guide the optimisation. The proposed algorithm is implemented in C. A new design model is proposed to implement partitioned sub-FSMs, which makes the existing monolithic FSM state assignment can be applied to partitioned FSMs. The experiment results show that the proposed approach can reduce power dissipation up to 78%.