Cluster randomized trials (CRTs) are experiments in which clusters of persons, rather than the persons themselves, are randomized to receive one of the interventions being studied. The use of CRTs has been increasing in response to the attention being paid to pragmatic as opposed to explanatory clinical trials, comparative effectiveness research, and community health promotional activities. We describe and illustrate the use of CRTs in these and other applications. Special attention is paid to ethical challenges in the design of such studies, and to tools for facilitating the implementation of interventions found to be efficacious in the trial into everyday clinical practice or effective community-wide programs. We argue that while CRTs have many useful and valid applications, there can be times when their use should be precluded due to ethical constraints. Special vigilance is required in research carried out in developing countries, where villages often seem to be a natural choice for clusters, but considerations of 'standard of care' may lead to control villages receiving no care or services. Full-fledged randomized controlled trials are not required to show that people who are doing poorly because of living in squalid conditions without proper sanitation and health care will, in the absence of change, continue to do so.