Vehicular Networked Systems are Cyber-Physical Systems that integrate vehicular com- munication, control and information processing technologies to achieve more safe and ef- fective operation of autonomous vehicles. Active coordination between vehicles is often used to improve system safety. This coordination may be achieved by exchanging informa- tion over wireless communication networks. Wireless communication networks, however, have limited reliability since they may drop bursts of information packets. This variation in channel state, moreover, may be a function of the vehicle's physical state. One of the great challenges faced by self-driving vehicles is finding ways of assuring system safety and efficiency in the presence of such unreliable communication channels. The first part of this dissertation addresses safety issues. We first propose a novel chan- nel model that we call the state dependent exponential bounded burstiness or SD-EBB channel model. The SD-EBB channel captures the burstiness and the channel's depen- dence on the vehicle's physical state. We show that the SD-EBB channel is more general in the sense that it can characterize i.i.d. channels as well as two-state Markov chain chan- nels. This dissertation uses the SD-EBB channel to develop a distributed switching control strategy that enforces an almost-sure notion of system safety. Almost sure stability requires that the likelihood of the vehicular states entering a pre-defined safe region asymptotically goes to one as time goes to infinity. Necessary and sufficient conditions are presented to  enforce almost sure safety for a nonlinear cascaded vehicular system. The proposed dis- tributed switching strategy is then applied to formation control of a leader-follower chain. Experimental results show that the system achieves almost sure safety under the switching strategy and is more safe than the non-switching scheme. When the vehicular networks have limited bandwidth, we develop a novel event triggering scheme under which the sys- tem is almost surely safe in the presence of deep fades. Furthermore, we also show that the proposed event triggering scheme reduces the transmission frequency as system states approach the safe set thereby providing an efficient use of network bandwidth. This is in contrast to traditional event triggering schemes where Zeno phenomena may occur due to deep fades. The second part of this dissertation proposes a co-design paradigm to ensure both vehi- cles' safety and efficient use of transmission power for vehicular networked systems. This co-design paradigm is formulated as a constrained cooperative stochastic game where the equilibrium points represent the optimal control and power strategies to achieve system safety and efficiency. We show that the equilibria of the stochastic game are equivalent to the solutions of a generalized geometrical programming problem. Since the generalized geometric programs are usually non-convex, two relaxed convex geometric programming problems are formulated to provide upper and lower bounds on the optimal solutions. A branch-bound algorithm is employed to asymptotically approach the optimal solutions.