This paper constructs an efficient, budget-balanced, Bayesian incentive-compatible mechanism for a general dynamic environment with quasi linear payoffs in which agents observe private information and decisions are made over countably many periods. First, under the assumption of “private values” (other agents’ private information does not directly affect an agent’s payoffs), we construct an efficient, ex post incentive-compatible mechanism, which is not budget-balanced. Second, under the assumption of “independent types” (the distribution of each agent’s private information is not directly affected by other agents’ private information), we show how the budget can be balanced without compromising agents’ incentives. Finally, we show that the mechanism can be made self-enforcing when agents are sufficiently patient and the induced stochastic process over types is an ergodic finite Markov chain.