In the 30 years since the Belmont Report, the role of the community in research has evolved and has taken on greater moral significance. Today, more and more translational research is being performed with the active engagement of individuals and communities rather than merely upon them. This engagement requires a critical examination of the range of risks that may arise when communities become partners in research. In attempting to provide such an examination, one must distinguish between established communities (groups that have their own organizational structure and leadership and exist regardless of the research) and unstructured groups (groups that may exist because of a shared trait but do not have defined leadership or internal cohesiveness). In order to participate in research as a community, unstructured groups must develop structure either by external means (by partnering with a Community-Based Organization) or by internal means (by empowering the group to organize and establish structure and leadership). When groups participate in research, one must consider risks to well-being due to process and outcomes. These risks may occur to the individual qua individual, but there are also risks that occur to the individual qua member of a group and also risks that occur to the group qua group. There are also risks to agency, both to the individual and the group. A 3-by-3 grid including 3 categories of risks (risks to well-being secondary to process, risks to well-being secondary to outcome and risks to agency) must be evaluated against the 3 distinct agents: individuals as individual participants, individuals as members of a group (both as participants and as nonparticipants) and to communities as a whole. This new framework for exploring the risks in community-engaged research can help academic researchers and community partners ensure the mutual respect that community-engaged research requires.