Corruption is an important social and ethical problem / fight with it requires changes in values, norms and behavioral patterns of the society. This is usually a long and difficult process. Decades should pass to change deep values of a society. In the mean time, it is possible to combat corruption by changing incentive structures in the economy. If deep causes of the problem are analyzed carefully, a new system of governance can be established, such that, even most opportunist individuals do not find getting involved in corrupt practices profitable. Aim of this thesis is to examine characteristics of the system providing a fertile environment for corruption and to figure out factors stimulating corrupt transactions using game theoretical models. The first two models examine corruption as a kind of transaction between the briber and the bribee. In the models, it is shown that intermediaries sector occur from the profit maximization behavior of agents. This sector, by establishing long term, trust based relationships with bureaucrats, decreases risks occurring from the fact that the two parties involved in a corrupt transaction do not know each other perfectly. This sector, by reducing the likelihood of detection, serves corrupt transactions, and in return for the service it provided, takes commission, so gets benefit. Third model examines a strange type of corruption, a case of (spurious) middlemen obtaining bribe from the public service bureaucrats give, by pretending that he has influence on the acceptance or speed of it. The model tries to detect the characteristics of the environment making such a deception process persistent.