The concept of need is often proposed as providing an additional or alternative criterion to cost-effectiveness in making allocation decisions in health care. If it is to be of practical value it must be sufficiently precisely characterized to be useful to decision makers. This will require both an account of how degree of need for an intervention is to be determined and a prioritization rule that clarifies how degree of need and the cost of the intervention interact in determining the relative priority of the intervention. Three common features of health care interventions must be accommodated in a comprehensive theory of need: the probabilistic nature of prognosis (with and without the intervention); the time course of effects; and the fact that the most effective treatments often combine more than one intervention. These common features are problematic for the concept of need. We outline various approaches to prioritization on the basis of need and argue that some approaches are more promising than others.