Monetary policy is now mainly about setting short-term interest rates, despite all the well-known difficulties that choosing the “correct” interest rate entails. This article follows an alternative line of reasoning, for which there is an overwhelming theoretical case. There has been a major measurement error in virtually all of the previous literature on money. Instability in empirical relationships has been primarily due to the fact that simple-sum measures of money are not admissible aggregates on index-theoretic grounds. This error has been especially important in a period when characteristics of components which are added together have been changing. The authors do not claim that correction of this measurement error salvages entirely the role of money as a macroeconomic indicator (though such may still be the case). Rather, their primary focus is to see whether acceptable indexes of money outperform traditional money measures in conventional tests. As is often the case in applied studies, the evidence turns out to be mixed but leaning in favor of the superiority of weighted over simple-sum aggregates.