The optimization of coordinated recovery of units and important loads after blackout is studied. Since the contribution to the power system recovery process of different units in the same plant is neglected in former studies, a coordinated two-layer unit restarting considering important load recovery optimization model is established in this paper. The units to be recovered are divided into network-layer units and plant-layer units. By studying the interaction of two-layer units and important loads, the coordinated multi-objective hierarchical restoration method for units is proposed. The continuous unit restoration process is divided into sequential time steps. For each time step, the non-dominated sorting genetic algorithm and the combined use of CRITIC objective weighting method and grey relation projection method are employed to solve the unit restoration optimization problem. Furthermore, the fuzzy analytical hierarchy process and the greedy algorithm are employed to determine the restoration scheme of important loads for each time step. The effectiveness of the proposed method is validated by the optimization results on the New England 10-unit 39-bus power system. This work is supported by National Natural Science Foundation of China (No. 51277076) and Fundamental Research Funds for the Central Universities (No. 2014XS75).