This commentary takes up conceptual issues raised in the introduction. Discussing the contributions to this volume, we offer three points. First, the concept of fragmentation needs theoretical clarification, which can be provided to some extent by sociological differentiation theory. We suggest a typology of different types of fragmentation. Second, differentiation theory helps to improve understanding of the different causes of fragmentation. Third, a high level of institutional differentiation is an important characteristic of modernity at many levels of politics. It is not fragmentation per se, but rather the (lack of) coordination of fragmented or differentiated institutions, that is a problem for global governance.

One of the most striking features of the work of scientists specialized in regional environmental governance is the huge diversity of ways they refer to the notion of region. In this academic subfield, “regionality” refers to different orders of reality (ontology), and regions have a heterogeneous status in the production of knowledge (epistemology). While such a diversity of uses and meanings illustrates the rich potential of a regional scope in environmental governance analysis, scholars' ontological and epistemological stances must be made more explicit. The objective of this commentary is to elaborate this suggestion and to illustrate it on the basis of the articles published in this special issue.

The Oslo-Potsdam approach to regime effectiveness has much to recommend it, especially in conceptual terms. It yields effectiveness scores for individual regimes that range from 0 to 1 and that facilitate comparative analysis. Yet the revised version of the approach set forth in Hovi et al. (2003, this volume) fails to solve the fundamental problems evident in earlier versions regarding the no-regime outcome (NR) and the collective optimum (CO). Nor does this version address the relative merits of the Oslo-Potsdam solution and other approaches to regime consequences that do not rely on direct measurements of regime effectiveness as the dependent variable. As a result, the argument that some measure of effectiveness—however faulty—is better than none is not persuasive. Even so, the debate over the Oslo-Potsdam solution has proven fruitful. Our understanding of the issues involved in evaluating regime consequences has surely grown as a product of this debate.