Research on policy communities, policy networks, and advocacy coalitions represents the most recent effort by policy scholars in North America and Europe to meaningfully describe and explain the complex, dynamic policy making processes of modern societies. While work in this tradition has been extraordinarily productive, issues of collective action have not been carefully addressed. Focusing on the advocacy coalitions (AC) framework developed by Sabatier (1988) and Sabatier and Jenkins-Smith (1993) as an example of a productive research program within the policy network tradition, this article (1) examines the potential of the AC framework, with its emphasis on beliefs, policy learning, and preference formation, to provide richer explanations of policy making processes than frameworks grounded exclusively in instrumental rationality; (2) suggests that paradoxically, however, the AC framework can more fully realize its potential by admitting the explanations of collective action from frameworks based on instrumental rationality; (3) incorporates within the AC framework accounts of how coalitions form and maintain themselves over time and of the types of strategies coalitions are likely to adopt to pursue their policy goals; and (4) derives falsifiable collective action hypotheses that can be empirically tested to determine whether incorporating theories of collective action within the AC framework represents a positive, rather than a degenerative, expansion of the AC framework.