Experimental studies in monetary economics usually study infinite horizon models. Yet, the time constraints of the laboratory sessions in which these models are conducted create finite horizons that imply monetary equilibria cannot exist. Moreover, laboratory subjects do not treat the probabilistic termination rule typically used in a manner consistent with the discount factor that the rule is intended to replace. Thus, it is unclear whether these experiments evaluate subjects' use of money to ameliorate trading frictions as an equilibrium phenomenon, their inability to understand backward induction, or features of games that promote the use of money behaviorally, even when doing so is not an equilibrium strategy. To address this issue, we present a pair of finite-horizon games where monetary exchange is an equilibrium, and report an experiment that evaluates behavior in these games in light of a finitely repeated alternative where monetary exchange is not an equilibrium.