The aim of this paper is to reflect on the problem of feudal dynamics on the basis of an empirical study. The existence of processes of social differentiation and the chances of such processes developing further are comparatively analysed in the context of three different types of Castilian lordship that implied distinctive political configurations. The operative nucleus of W. G. Runciman’s theory is applied to the empirical evidence, in that the concept of ‘mode of the distribution of power’ permits us to deal with political power as a structural element, while the notions of ‘systact’ and of inter-societal ‘competitive selection of practices’ permit us to carry out a systematic comparative analysis and to establish a link between structure and dynamics. Through the analysis of social practice, it is shown that accumulative tendencies can be found in a pre-capitalist context, although they were contingent and variable according to the way in which political power was exercised.