This study evaluates the limitations and relevance of non-cooperative games as a tool for analyzing political and policy behavior. It aims to answer the questions, “for what situations are such non-cooperative games appropriate, and for what do they fail to predict actual behavior?” To answer these questions, we are going to look into the assumptions underlying the existence of equilibria developed by this non-cooperative game model. This model uses two types of situations: extensive form and normal form. Each of them is evaluated in terms of its underlying assumptions and their relevance to real behaviors in the public sector. The several theorems concerning non-cooperative game are introduced and evaluated in terms of its meaning and relevance to the real public arena. In particular, we see that the zero sum assumption is a restriction that precludes modeling many other situations central to politics and public administration, and that the prisoner dilemma models important processes, including collective action problem.