SUMMARY The definition of causalgia as a pain state following peripheral nerve injury has been accepted since the term was introduced by Weir Mitchell over a century ago. In the present paper, problems of nomenclature and nosology are discussed, and attention is drawn to the fact that the same clinical features can occur spontaneously, in nontraumatic nerve lesions, in the absence of a part as in phantom limb states, and in diseases confined to the central nervous system. Attention is also drawn to the lack of correlation of pain with the effects mediated by catecholamines in the sympathetic nervous system and with the response to sympathetic blockade. Concerning mechanisms, a number of peripheral mechanisms have been postulated. These are reviewed, and while they might be correct when causalgia arises from peripheral nerve damage, they cannot provide adequate explanation for at least some instances of causalgia. The relevance of the neuroma as a model for chronic pain in general, and causalgia, is questioned. Also questioned is the view that causalgia is a state that depends on peripheral involvement of the sympathetic nerve supply. Certain authors in the past considered that the central nervous system (CNS) played an important part in causalgia, and current evidence supporting this view is assessed. Involvement of the CNS is suggested by (1) the development of causalgia in diseases confined to the CNS and in phantom pain states; (2) the unusual distribution of pain sometimes experienced; (3) the paradoxical development of widespread pain that can occur after damage to the sympathetic nervous system; (4) the effects of peripheral sympathetic blockade even when the cause lies centrally; and (5) central interactions with motor, sensory and psychological phenomena Reservations concerning the role of catecholamines in causalgia are outlined, and the possibility is considered that nonadrenergic substances may be implicated.