This paper uses multi-period cross-sectional data on financial assets holdings to shed light on the postwar stability of money demand in the United States. I first present a new measure of the evolution of financial market participation, by relating participation to the extensive margins of money demand, and quantify the influence of wealth on participation decisions. I then relate the increase in participation to the period of "missing money" and to the subsequent higher interest rate elasticity of monetary aggregates. The paper indicates that time series estimations of money demand relationships are inherently flawed and tend to inappropriately suggest instability.