The article reviews a family of multilevel models that can be used to build general theories of the nonprofit sector that are still sensitive to variations in context. The comparative study of the nonprofit (or nongovernmental) sector presents formidable challenges to social scientists who are attempting to advance theory on the sector. Ostensibly, the goal is to model and test theories that are generalizable. Yet, as scholars study topics such as volunteerism, donations, governance, management, advocacy, accountability, and the like in different political, economic, and cultural contexts, they often find different patterns across cases. After reviewing the issues and introducing the idea that time (or more specifically events) can be thought of as context as well, we present an analytical approach for doing comparative research using the framework of hierarchical linear modeling.