In this dissertation, I study strategic communication, in which a sender strategically discloses information to persuade a receiver to take an action favorable to the sender. I study two models of constrained communication. The first one generalizes the standard Bayesian Persuasion model to allow for the receiver's strategic behavior. The second one, joint work with Nour Chalhoub, studies a dynamic information disclosure under the assumption that the sender can only lie in one direction, by under-reporting the state, but never over-reporting it. The constraints in each model are intuitive for real-life application and lead to results that are of substantial difference from the results of the standard settings.