The growing problems in water pollution at a watershed scale have caused more and more conflicts between upstream and downstream regions, making ecological compensation a vital approach to balance the interests of all stakeholders within the watershed concerned. Presently, ecological compensation has been practiced in some areas and gained certain progress, but problems have been revealed during the practice as well. In this paper we tend to illuminate the importance of system arrangement by analyzing the changes in benefits of stakeholders in upstream and downstream under various systems. We introduce the basic concepts involved in explaining the connotation of the systems. We then compare two cases (Prisoner's Dilemma and Prisoner's Dream) to better illustrate how efficiency is influenced by the systems. Based on the case studies, we point out that efficiency and equity are determined by the current systems, and the meaning of efficiency and equity might be different in each system. We also analyze the role of the systems in promoting "efficiency and equity"--which is narrowly defined in this paper--of watershed ecological compensation based on a hypothetical example. In this example, A represents the factory upstream while B represents the resident downstream being polluted by the waste water from factory. Besides, There are four types of system in this example: In the first system, A does not have to take into account the loss of B, and their total income has realized the maximization, but the majority of people will think this system is unfair to B; In the last three "fairer" systems, each has a compensation plan for B, the income of both sides also has reached the maximization in the third system as the same as it in the first system, while the total income in the fourth is the lowest of all in which A is required to install sewage treatment facilities. We conclude that (1) it will probably gain both efficiency and equity by designing a reasonable system and (2) giving stakeholders more power to affect institutional design will promote the social equity. Our results could provide support to building an analytical framework for decision makers and has significant values for alleviating the conflict of interest between upstream and downstream stakeholders in the practice of watershed eco-compensation.