Although many decisions involve a stream of payoffs over time, political scientists have given little attention to how actors make the required tradeoffs between present and future payoffs, other than applying the standard exponential discounting model from economics. After summarizing the basic discounting model, we identify some of its leading behavioral anomalies--declining discount rates; preference reversals; higher discount rates for smaller payoffs than for larger payoffs and for gains than for losses; framing effects based on expectations; and a preference for ascending rather than descending sequences. We examine the leading alternative models of discounting and then apply a quasi-hyperbolic discount model to the problem of cooperation in iterated Prisoner's Dilemma games. We demonstrate that if actors display the widely observed tendency to highly discount the immediate future, then cooperation in an iterated Prisoner's Dilemma game is more difficult than Axelrod suggests. (Contains 4 figures and 26 notes.)