Following the Treasury-Federal Reserve Accord of March 3, 1951, the Federal Open Market Committee (FOMC) focused on free reserves - the difference between excess reserves (reserve deposits in excess of reserve requirements) and borrowed reserves - as the touchstone of U.S. monetary policy. However, managing free reserves was problematic because highly variable and not readily predictable autonomous factors, including float, Treasury balances at Federal Reserve Banks, and currency in the hands of the public, induced comparable volatility and unpredictability in reserve deposits and hence in free reserves. Managing free reserves effectively required policy instruments that could inject and drain large quantities of reserves quickly at low transaction costs. This paper surveys the two leading policy instruments for reserves management: 1) open market purchases and sales of Treasury bills, and 2) repurchase agreements. Outright transactions in bills were specifically authorized by statute and used in unexceptional ways for managing reserves over relatively long periods, but they had significant drawbacks for short-term "in and out" operations when additional reserves were needed for only a few days. Repos, however, while not specifically authorized by statute, were ideally suited for in-and-out operations. The acceptance of repurchase agreements as an instrument of monetary policy, even in the face of active resistance by some FOMC members, illustrates how utility can sometimes trump concerns about statutory authority, equity, and need.