The legal terms "benefit" and "added benefit" and the procedures related to their assessment, given the application of a new medical intervention for approval, should help to restrict the market to those products for which there is much evidence of their benefit or added benefit. The term implies the fiction of an overall benefit for all patients with the same disease. However, from the perspective of health services research and that of epidemiology the term inevitably has to be extended to cover the benefit to a variety of groups of patients or users of a broad spectrum of medical interventions in the real world, ranging from inpatient treatment to vaccination or screening programs. Thus, the assessment of benefit requires a comparison of the new product with all the alternatives currently available for routine care, explicitly taking into account user preferences. Hence, the assessment of benefit in health services research is not one-dimensional and requires new types of studies that go beyond the traditional phase III trials (efficacy trials). New approaches are mainly developed by comparative effectiveness research (CER). CER studies also use randomized designs, because they are currently the best available method for causal inferences. However, randomization in CER is extended to a much broader framework, including health-related databases, registers, epidemiological studies, feasibility studies, and post hoc analyses. CER has developed the necessary and appropriate designs and statistical methods. In addition, some of these methods allow an adaptive assessment of benefit, which can be used to monitor the implementation of new health care interventions or programs.