In a wireless cooperative network, system reliability can be improved by introducing network coding (NC) for transmitting data packets from user to destination through relay nodes. At the destination, a decoding strategy is required to recover the original data packets. The use of NC in cooperative networks has been intensively studied in previous works in terms of the conventional model for two users and a single relay in a network. However, the network model cannot act as a virtual multiple-input multiple-output system, and a multi-user multi-relay network model could be used in a real system. Therefore, this paper proposes an improved model of two network decoding strategies, selection with soft combining (SSC) and selection with hard combining (SHC), for multi-user multi-relay cooperative networks. Users are classified based on their channel conditions, with better signal-to-noise (SNR) ratio sources being viewed as strong users, and others as weak or moderate users in the decoding strategies. To evaluate the performance of the proposed model, we first derive the bit error probability expressions for each strategy as a function of SNR and then evaluate the performance using numerical simulation for a Rayleigh fading channel. Simulation results show that SSC outperforms SHC. Furthermore, the improvement in network performance is achieved either by having a higher modulation level or using incremental relaying as the signal reception method at the destination.