This paper clarifies some basic concepts or assumptions of the prisoner's dilemma, asserts the independence between the two agents A and B, and advocates the application of the dominance principle of decision theory to the prisoner's dilemma. It discusses several versions of the prisoner's dilemma, including the one-shot and repeated cases of a noncooperative game from a purely egoistic point of view. The main part of this paper, however, is a study of the problem from a moral point of view through a special decision-theoretic approach. Morality is taken into account by incorporating the utility of the feeling of moral satisfaction for the agent, as a part of the total utility for the agent, into the decision-theoretic model. In this way the problem will appear as a purely technical decision problem, and the conflicts between various assumptions, or the dilemma caused by the problem, will no longer exist. It is also pointed out that in a more general case, for some values of the coefficient of morality k, dominance will not exist so that the dominance principle will not be applicable.