That cooperation is at best difficult and at worst impossible in the anarchic international system is an established idea. Analyses usually assume that state preferences put them in a prisoner's dilemma where defection is the strictly dominant strategy. Explaining cooperation means showing why actors forgo the short-term benefits of exploiting others. Mechanisms involve conditional play, institutional arrangements, and hegemonic leadership, which help overcome coordination, monitoring, enforcement, and free-rider problems.

Andrew Kydd argues that trust can play no role in such a world because cooperation is based on fear of being punished, rather than on a belief that it would not be exploited. Kydd defines trust as one's subjective belief that the other actor prefers to reciprocate cooperation rather than exploit it. While mistrust figures prominently in realist theories (e.g., causing escalatory spirals under the security dilemma), they have been vague about its origins, impact, and the ways actors can overcome it. To analyze these issues, the author uses a series of game-theoretic models in which actors are uncertain whether the other player is security seeking or expansionist. Trust is the minimum belief that the other is security seeking to the extent necessary for a security seeker to cooperate. Cooperation is possible when the two actors sufficiently trust each other.

The bulk of the book comprises three parts, each consisting of a theory chapter, which develops the formal argument, and an empirical chapter, which examines the historical record in light of theoretical implications. The empirical chapters are not meant as tests of the theory. Because any pattern of behavior can occur in some equilibrium, any observable event can be rationalized with a set of beliefs and preferences. Since many equilibria are not mutually exclusive, the fact that an equilibrium pattern fits the historical record provides only tenuous support for the theory. Instead, Kydd uses the theory as a guide to the type of evidence one would be most likely to find if the model correctly captured a significant part of real world interaction.

Part II shows that in a model where actors are uncertain about each other's beliefs, rational play can result in both tragic and nontragic spirals, but the latter are far more likely. The implication helps explain the origins of the Cold War. Whereas revisionists insist that the Cold War was essentially a tragic spiral between two security-seeking rivals, Kydd argues that the nontragic traditionalist account, which pits a security-minded and initially trusting United States against a potentially expansionist and suspicious Soviet Union, was fundamentally correct.

Part III extends the trust argument to multilateral settings. The model shows that hegemony can lead to cooperation, but only if the leading state is sufficiently trustworthy. Kydd then examines the interaction between Western European states and the United States in the first decade of the Cold War on key questions of defense, and German partitioning and rearmament. He argues that to understand the period, one must account for the steps the Americans took to reassure their European allies. Although a bit vague at times about what precisely the Western Europeans were fearful of (American abandonment? German revanchism? Soviet expansionism? all of the above?), Kydd effectively counters both alternatives of benign hegemony (the United States provided public goods unilaterally) and coercive hegemony (the United States forced contributions).

Part IV explains how players can overcome mistrust by making costly gestures to signal their preferences and shows how the Soviets were able to reassure the Americans that the USSR was no longer expansionist in the late 1980s. The Reassurance Game shows that a security seeker can take risks by cooperating while being unsure about the other actor. Since this behavior separates him or her from an expansionist, it builds trust that leads to more cooperation, provided the other side is similarly motivated. Mikhail Gorbachev's strategy was to act in a way that an expansionist leader would not have. This is why the series of dramatic, often unilateral, steps he took to disengage from Europe eventually persuaded even the most suspicious observers not to fear the Soviet Union. As in any game-theoretic account, the book is agnostic on the sources of preference formation and hence cannot speak to why this change occurred.

Three modeling choices suggest avenues for future work. First, all models restrict players to two actions: cooperation or defection. If an action can reveal something about preferences, information transmission can happen only when actors want it to and are able to play separating strategies that other types would be unwilling or unable to mimic. Within the limited action space, players cannot tailor behavior to signal preferences or elicit from others actions that would reveal theirs. Second, since trust is about beliefs, revising these beliefs is crucially important. Except for the Reassurance Game, learning is strategically irrelevant because (a) the games end before actors can act on their updated beliefs, or (b) new information comes from nonstrategic reports rather than the opponent's behavior. We cannot use such models to study bluffing, probing, and other forms of strategic misrepresentation that seem relevant. Third, even though most of the analysis is couched in dynamic terms, only the Reassurance Game embeds a meaningful time dimension in that players can update their beliefs and then act on the new information. (It is not even clear how one can think of spirals in a static setting.) One is left wondering why this game was not used as the basis for the entire book.

Trust and Mistrust in International Relations is an important book that does what a good theory book should do: offer a novel way of looking at a phenomenon and provide a solid basis for future research. It is an admirable piece of work because it develops the main ideas with logical rigor without sacrificing readability, a rare thing indeed in the often highly technical world of formal modeling. But perhaps most impressively, Kydd utilizes the models to illuminate crucial periods of recent history in a way that allows one to weigh, and dismiss, well-known alternatives.