We study joint end-to-end congestion control and per-link medium access control (MAC) in ad-hoc wireless networks. We use a network utility maximization formulation, in which by adjusting the types of utility functions, we can accommodate multi-class services as well as exploit the tradeoff between efficiency and fairness of resource allocation. Despite the inherent difficulties of non-convexity and non-separability of the optimization problem, we show that, under readily-verifiable sufficient conditions, we can develop a distributed algorithm that converges to the globally and jointly optimal rate allocation and persistence probabilities. A key contribution is that our results can accommodate general concave utility function rather than just the logarithmic utility function in existing results. © 2006 IEEE.