An increasing number of monetary economists now describe the money supply process within the framework of the Brunner-Meltzer nonlinear hypothesis. For a detailed description of the hypothesis see [4, 5, 6]. Several aspects of this general approach to analyzing the behavior of the money stock are considered in this article. In section 1, the relative merits of different ways of decomposing the money supply into a monetary multiplier and a base are discussed. The difficult task of separating the effects of reserve requirement changes from the monetary multiplier is discussed in section 2. A problem with Brunner and Meltzer's reserve adjustment variable is pointed out and an alternative logarithmic reserve adjustment variable is suggested. These two reserve adjustment variables are then compared in section 3. In section 4, the behavior of the monetary multiplier during the period 1961-74 is examined and an explanation is presented for the marked increase in the variance of the monetary multiplier. The final section considers some simple forecasting models that utilize the basic framework of the Brunner-Meltzer nonlinear hypothesis. It is shown that, despite the large variance in the monetary multiplier, it is possible for the Federal Reserve to control the money supply quite accurately over periods of one or two quarters.