Device-to-device (D2D) communications are now considered as an integral part of future 5G networks which will enable direct communication between user equipment (UE) without unnecessary routing via the network infrastructure. This architecture will result in higher throughputs than conventional cellular networks, but with the increased potential for co-channel interference induced by randomly located cellular and D2D UEs. The physical channels which constitute D2D communications can be expected to be complex in nature, experiencing both line-of-sight (LOS) and non-LOS (NLOS) conditions across closely located D2D pairs. As well as this, given the diverse range of operating environments, they may also be subject to clustering of the scattered multipath contribution, i.e., propagation characteristics which are quite dissimilar to conventional Rayeligh fading environments. To address these challenges, we consider two recently proposed generalized fading models, namely $κ-μ$ and $η-μ$, to characterize the fading behavior in D2D communications. Together, these models encompass many of the most widely encountered and utilized fading models in the literature such as Rayleigh, Rice (Nakagami-$n$), Nakagami-$m$, Hoyt (Nakagami-$q$) and One-Sided Gaussian. Using stochastic geometry we evaluate the rate and bit error probability of D2D networks under generalized fading conditions. Based on the analytical results, we present new insights into the trade-offs between the reliability, rate, and mode selection under realistic operating conditions. Our results suggest that D2D mode achieves higher rates over cellular link at the expense of a higher bit error probability. Through numerical evaluations, we also investigate the performance gains of D2D networks and demonstrate their superiority over traditional cellular networks.