As we approach the new millennium, it is clear that we are on the brink of a major change in clinical pain management. We are poised to move from a treatment paradigm that has been almost entirely empirical to one that will be derived from an understanding of the actual mechanisms involved in the pathogenesis of pain. When this is achieved, pain treatment will at last be rationally based. The implications of this are immense and will necessitate major changes in the way we classify pain, which until now has been based on disease, duration and anatomy, to a mechanism-based classification. In addition, the assessment, diagnosis and treatment of pain will change. The aim in the future will be to identify in individual patients what mechanisms are responsible for their pain and to target treatment specifically at those mechanisms. We present for discussion, a new approach for classifying pain, based on an analysis of mechanisms, and show how this could be used to assess pain clinically. Such kinds of pain assessment, which need to be designed to reveal as much as possible about mechanisms, are necessary for more sophisticated epidemiology and clinical research as well as for providing the outcome measures necessary for the evaluation of the efficacy of new treatments targeted at particular pain mechanisms.