Abstract-In this paper, we propose Unequal Error Protection (UEP) coding theory as a viable and flexible method for the design of network codes for multi-source multi-relay cooperative networks. As opposed to state-of-the-art solutions available for improving the diversity gain of cooperative networks, it is shown that the proposed method allows us to assign each source node the desired diversity gain, according to, e.g., the requested Qualityof-Service (QoS) or power constraints. The diversity advantage of the UEP-based network code design over conventional relay-only and XOR-only solutions is shown for the canonical two-source two-relay network. Furthermore, Maximum-Likelihood (ML-) optimum channel-aware receivers for multi-source multi-relay cooperative networks are developed, and their Average Bit Error Probability (ABEP) and achievable diversity over fading channels analytically studied. It is shown that only a cross-layer (joint) implementation of de-modulation and network-decoding allows the destination to fully exploit the diversity inherently provided by the distributed network code. Finally, analytical derivations and findings are substantiated via Monte Carlo simulations.