In this paper, we address the robust source and relay matrices design for the multiple-input multiple-output (MIMO) relaying broadcast channels (BC) with imperfect channel state information at the transmitter (CSIT). Our objective is to maximize the minimum achievable rate among all users, which dominates the quality of service (QoS) performance of the system. In the proposed scheme, we first set up an equivalent problem, and then relax the constraints of the new problem to decouple it into three tractable subproblems. Finally, an iterative algorithm is proposed to jointly optimize the source and relay matrices. The advantage of the proposed scheme is demonstrated by numerical experiments.