This article attempts to draw together into a coherent descriptive theory a number of ideas which have been developed in recent writing on the ways in which the parties to a conflict interact with each other and with the normative and institutional structures available to them (“litigation behavior”). A review of the literature leads to the formulation of a number of minimum requirements which any such theory must satisfy; among other things, the concept of a “case” as the unit of analysis of such a theory is shown to be radically inadequate, and the necessity of studying litigation behavior in a context conceived of as legally pluralistic is emphasized – the article makes, in this latter connection, a special effort to integrate the anthropological with the sociological tradition in the study of litigation. Next, it is argued that the unit of analysis of litigation theory should be “normative claims” (rather, for example, than “conflict”), but that the “baseline” in terms of which the emergence and progress of such claims should be measured must be sought not, as is generally the case in the litigation literature, simply in 100 % of the unit of analysis itself (since the occurence of a “claim” is itself a problematic event), but rather in something which is not itself subject to the influence of the very same variables which are associated with the progress of litigation: “relationships” of the relevant sort are suggested as the appropriate baseline for litigation research. There follows a distinguishing of three aspects of an overall theory – litigation structure, litigation system output, and litigation process. After a few Observations concerning the first two, the remainder of the article is devoted to litigation process. Existing more or less implicit maps of litigation careers are examined and found wanting; an alternative map is proposed which (as required if one takes account of the legally pluralistic character of litigation) locates litigation in the context of the various “semi-autonomous social fields” which constitute society. The moments of actor-choice identified by such a map are associated with three groups of factors: characteristics of litigation structure, of the actors, and of their relationships. In particular, the last of these is extensively discussed. The article concludes with some discussion of the weak points in the litigation literature as revealed by an attempt to construct a coherent theory, and a recapitulation of the main themes of the article.