This study investigates the impact of imperfect channel state information that is caused by feedback delay with time-variant channels and also by channel estimation errors with an opportunistic source-pair selection strategy. The investigation considers multi-user bi-directional wireless relaying networks. The opportunistic source-pair selection strategy selects the appropriate source-pair node among N available nodes. The selection strategy is based on each source pair's maximum sum-rate in each time slot. According to analytical, asymptotic, and Monte-Carlo simulation results, the opportunistic source-pair selection strategy, besides achieving a near-optimal solution for inter-cell interference that is caused by other user-pairs in the system model, also achieves diversity order in high signal-to-noise regimes. Moreover, the opportunistic source-pair selection strategy outperforms the max-min-based selection strategy in such a system model. Results also show that feedback delay degrades the achievable diversity order from N to 0 while affecting coding gain. The results also show that the channel estimation error case does not affect the achievable diversity order, but it does affect the system coding gain in low-signal-to-noise regimes. Moreover, the channel estimation error effects upon system coding gain become negligible at high-signal-to-noise regimes.