For a decode-and-forward (DF)-based multiple-input-multiple-output (MIMO) relaying system, we propose a novel partial-information relaying scheme that adaptively controls the number of streams to be forwarded to the destination. In the first transmission phase, a source node broadcasts data streams that consist of nonforwarding and forwarding streams. In the second transmission phase, a relay node forwards only forwarding streams to the destination, and a destination node decodes both nonforwarding and forwarding streams by successive interference cancellation. We consider two possible scenarios according to the availability of the channel-state information (CSI) at the source. The first scenario is that no CSI is available at the source, and therefore, a source node broadcasts data streams with equal power allocation. In this scenario, the number of data streams to be forwarded by the relay is determined according to the available CSI at the relay. For the second scenario, where the perfect CSI is available at the source, we propose a linearly combined precoding matrix for rate maximization. We also propose suboptimal schemes to reduce the complexity of the precoding design. Our analytical and numerical results show that the proposed scheme achieves a substantially higher rate than both the conventional relaying scheme, which forwards all the received streams, and the direct transmission scheme without relaying, regardless of CSI availability at the source.