This research thesis aims to add to the understanding of industrial distribution systems and to develop our ability to model such systems. The work is empirically based, taking a particular distribution system, studying it in depth and observing it during a period of change and reorganisation. The thesis is also part of a wider research programme to investigate the problems of applying Operation Research to decision making in unstable environments. Change in distribution systems often provides good examples of this form of decision making, and some tentative generalisations are drawn as a contribution to this wider programme. Empirical research into either distribution systems or decision making in unstable environments poses the problem of access. A characteristic of this form of decision making is its informality, without the usual stress on committees and reports found with more formal long range planning. In order to ensure access to the real decision making process the researcher became an indispensable part of that process. The research method adopted was of a type since become known as the dual researcher/change agent type. With this method the researcher had the two responsibilities of solving the particular problem while recording material as objectively as possible for research analysis. Hence a subobjective of the research programme was a project to diagnose and cure a problem arising in the distributive system under study. To do this two models were developed. Firstly a channel choice model which summarised the economic variables of the distributive system. By exploiting a particular structure inherent in the model, this could be recast into a form similar to the classical transportation algorithm. As the cost matrix was of a predominantly block-diagonal form an extended application of the saddle- point theorem allowed an efficient' dual decomposition procedure to be developed. The second model relied on attitudinal data, and attempted to model merchants’ behaviour in terms of how they construed their role in the distributive system and of the pressures placed upon them through the bargaining relationship. These models were used in tandem, the first demonstrating the ideal channel choice configuration from. the manufacturer's point of view, the second showing what could feasibly be achieved in spite of the power structure maintained by the merchants.