The idea of assessing needs both in individuals and in populations is popular in health and social care, but has serious conceptual shortcomings. The concept of needs does not distinguish between the identification of a problem and its solution. It inhibits a consideration of the probabilities as to how effective various interventions may be in any given case — nor does it reflect the iterative process that is the reality of most health and social care. It does not specify goals and oversimplifies evaluation of outcome because it does not take into account different degrees of change. In assessing population needs, there is the special risk of equating service use with service need, thereby entrenching the status quo. Instead of assessing needs, it is proposed that we identify problems, specify goals and choose interventions on the basis of probabilities of effectiveness. The outcome of any given intervention can be repeatedly reviewed with respect to its goals, and priorities may be reset accordingly.