Monitoring systems are designed to inform states whether their partners in cooperation are complying with their obligations or not. Under many conditions, many of which are elaborated by game-theoretic models of cooperation such as the repeated prisoners’ dilemma game (e.g., Axelrod 1984), information about compliance is crucial, and losing that information is indeed like losing the “holy grail.” Thus it is not surprising that mechanisms identifying compliance were once thought to be the raison d’être of international institutions.

In reality, the incidence of monitoring provisions is neither trivial nor universal; nor, when provisions exist, do they exhibit the same form. Consider the following:

1. • 60 percent of the agreements in the COIL random sample call for the creation of monitoring arrangements; 40 percent do not.

2. • Not a single monetary agreement in the COIL samples calls for the creation of a system of compliance monitoring.

3. • An astonishingly high majority (89 percent) of disarmament agreements call for a system of monitoring.

4. • 63 percent of human rights agreements delegate monitoring to a pre-existing IGO, but only 10 percent of economics agreements do.

This chapter explains both the incidence and form of monitoring systems. In a nutshell, agreements are particularly likely to call for a system to collect and disseminate information when there is substantial Uncertainty about the Behavior of other states, thereby corroborating Keohane's classic argument about the role of international institutions in the provision and dissemination of information.