Effective communication requires trust between the parties. However, situations with mistrust between communicating parties are commonplace: e.g., voters are often suspicious of politicians' statements because they worry that politicians are insincere; the court generally questions the credibility of witnesses; buyers might worry that a seller is not telling the truth in advertisements. In this thesis, I use game theory to study the role of trust in communication between a sender, who is the source of information about the state of the world, and an audience of receivers that make decisions based on the information provided. Formally, I capture trust, or lack thereof, between the sender and the audience as whether or not the sender can be trusted to follow a communication strategy. Thus, the sender's communication is Bayesian persuasion when there is trust (Kamenica and Gentzkow, 2011) and is cheap talk (Crawford and Sobel, 1982) when there is no trust. I focus on the case when the sender's preference is state independent so that the audience knows the sender's motives for providing information. The first chapter explores how the presence of diverse opinions in the audience can make the sender's cheap talk credible by making statements have stakes. Building on this intuition, I study how the sender can optimally communicate semi-publicly; i.e., by partitioning the audience into strategically formed groups and communicating publicly within each group but privately across groups. Effective semi-public communication involves the sender trading off the desire to tailor communication to individuals and the desire to gain credibility by ensuring appropriate diversity of opinions in the groups. Effective communication can be problematic because effectiveness is about whether the sender benefits, and not whether the audience benefits. In the second chapter, I study how the receiver can induce the sender to provide more information than the sender is otherwise willing by avoiding learning about the sender’s trustworthiness. Interpreting the receiver's investigations as cross-examination of witnesses in courts or audit of financial information, these results shed light on the importance of the investigator's incentives in enabling courts or buyers to obtain more information from strategic sources of information. The last chapter studies a two-player, undiscounted, infinitely repeated game in which only one player is informed about the state of the world, and players observe only each other's actions in each stage of the repeated game (Aumann, Maschler and Stearns, 1968). While players do not have explicit means of communication, the absence of discounting means that the informed player's actions in any initial finite stages of the game can be interpreted as costless messages sent by the sender in cheap-talk games. The main result of the chapter is a new characterisation of the informed player's equilibrium payoffs when the informed player's preferences are state independent.