A novel receding horizon control scheme is presented in this paper to manage networked leader-follower formation control problems in the presence of time-varying delay and data loss (measurement and feedback packet dropouts) occurrences on the communication and feedback channels. One of the main novelties of the proposed paradigm relies on the use of reachability analysis concepts involving the capability to combine sequences of predecessor sets with polyhedral piecewise state-space partitions. The predecessor sets are here derived offline by taking into account the prescribed state and input constraints, the interconnected nature of the system and the time delay induced by the network latency. They are used online to compute admissible and optimal control actions via a distributed receding horizon computational paradigm. Feasibility, uniformly ultimate boundedness and constraints satisfaction are formally proved regardless of any bounded time delays.