Multi-antenna relays can significantly increase the speed and reliability of wireless systems. However, because of the complexity of MIMO detection, there is considerable overhead in implementing a MIMO relay if the conventional detect-and-forward strategy is used. To address this challenge, we propose a novel cooperative partial detection (CPO) strategy that partitions the detection task between the relay and the destination. CPO leverages the structure of the tree-based c1ose-toML MIMO detectors, and modifies the tree traversal so that instead of visiting all the levels of the tree, only a subset of the levels, thus a subset of the transmitted streams, are visited. This novel approach reduces the tree levels, i.e. dimensions, in both the relay and the destination. Moreover, CPO provides a flexible method to control the level of partitioning between the relay and the destination, and thus, adjust the detection computational complexity in the relay and the destination. Monte-Carlo simulation results show that, under equal transmit power and complexity constraint in the destination, CPO achieves a better BER performance compared to the non-relay scenario, with limited computational overhead in the relay.