For wind-powered dc microgrids working autonomously, the available wind generation may be greater than the load demand, and hence, may cause supply-demand imbalance. To address this issue, this paper presents a new distributed load sharing control scheme, which coordinates the operation of multiple permanent magnet synchronous generators (PMSGs) in a dc microgrid. This scheme has a two-layer structure. Based on the distributed model predictive control approach, the upper layer controllers cooperatively optimize the operation of parallel-connected buck converters, which serve as the interfaces between PMSGs and microgrid bus. The optimized power references are sent to the lower layer wind generator controllers for execution. This way, the optimal load sharing among multiple PMSGs is achieved. The salient feature of the proposed scheme is that it optimizes the control actions in a distributed manner, which ensures that multiple PMSGs cooperate, rather than compete, with each other in achieving system-wide objectives. Simulation results verify the effectiveness of the proposed scheme.