This article explores some implications of adaptive learning for monetary evolution using a search-theoretic framework that allows for media-of-exchange network effects. Adaptive learning precludes any voluntary transition to a fiat standard from a non-monetary state of nature and can account for the historically-observed tendency for fiat monetary standards to emerge only following the prior appearance of commodity money and the widespread employment of redeemable banknotes. Adaptive learning can also account for governments' frequent resort to coercive measures to force a switch to fiat money and for their ability to affect such a switch even when doing so is not Pareto optimal. Exchange media fall into the general category of 'products for which the utility that a user of the good derives from [use] of the good increases with the number of other agents [using] the good' (Katz and Shapiro, 1985, p. 424). There are two reasons for these positive network effects. (1) Indirect. The more agents who employ a particular medium of exchange, the more likely it is that prices will be quoted in terms of that medium of exchange, i.e., that the medium of exchange will also serve as the economy's principle medium of account. (The costs of exchange are less for a medium of exchange that trades at par.) This indirect external effect is analogous to one encountered in computer hardware: the more purchasers of a particular kind of hardware, the more likely manufacturers of software are to supply compatible products. (2) Direct. The more persons who use a particular exchange medium, the better its quality or performance. This direct network effect is analogous to one encountered in communications technologies, such as telephones. The analogy is especially close for fiat money, which is useless to an isolated individual. This article examines some ways in which media-of-exchange network effects influence the evolution of monetary standards, and especially their role in determining both when and how a transition may occur from barter or commodity money to fiat money. The approach taken here emphasises the role of adaptive learning, where agents select exchange strategies that reflect static expectations concerning other agents' preferred strategies. Although the exchange game agents play generally has multiple Nash equilibria, including one in which all agents accept fiat money, adaptive learning precludes a spontaneous transition from barter to a fiat monetary standard even where a fiat standard would be welfare enhancing. Yet such learning does not rule out the spontaneous evolution of a commodity-money standard, or the eventual substitution of redeemable banknotes for commodity money. Readers familiar with the literature on network effects will recognise this