The paper offers an examination of the foreign exchange markets as they currently operate, followed by a consideration of the arguments over the desirability of a Tobin tax. Misgivings regarding how well the markets are working arise from recent apparent misalignments and crises, on the one hand, and from a set of seven academic findings, on the other hand. A review of the structure and size of the foreign exchange markets draws on the most recent central bank surveys released in September 1995. The BIS figure for the worldwide total is $1,230 billion of trading per day in April 1995. Importantly, less than one in five transactions is with a non-financial customer (.18 in London and New York). The case in favor of the Tobin tax features two major arguments. (1) Such a levy might reduce exchange rate volatility. A simple model giving this conclusion is presented in the Appendix. The starting point is a calculation showing that even a small tax would be a large disincentive to short-term transactions. The disincentive to long-term capital flows would be much smaller. This property does not extend to other forms of capital controls, and constitutes the beauty of the Tobin tax proposal. The crucial proposition then becomes that short-term speculation is on average destabilizing. Some support for this claim is cited, in the form of tests on survey data of exchange rate forecasts by market participants. (2) The Tobin tax would raise a lot of revenue more efficiently than alternative taxes such as tariffs. Some possible flaws in earlier estimates of revenue are pointed out here. The relevant base of transactions on which the tax would fall is larger than some have assumed, but the possible drop in trading volume in response to the tax is larger as well. A tax large enough to alter the structure of trading could conceivably collapse trading volume to as little as $151 billion/day. The author does not support a tax of this magnitude. Nevertheless, it is clear that even a more reasonable tax rate of 0.1 per cent would raise a lot of revenue, $166 billion per year in one estimate that is presented for the sake of concreteness. Whether this would be desirable depends heavily on the use to which the funds were put, or the alternative sources of tax revenue for which they are substituted. The case against the Tobin tax also has two major components. (1) It might create distortions and inefficiencies. As noted, it is even conceivable that the fundamental structure of the foreign exchange market could change, from decentralized and dealer-driven to centralized and customer-driven. Whether this would constitute a loss or gain in efficiency is difficult to say. (2) The Tobin tax would be extremely difficult to enforce. Along these lines lie the author's greatest doubts.