A microeconomic model of bank demand for borrowed reserves from the Federal Reserve is developed based upon constrained cost minimization. The derived demand function was found to correspond to behavior appropriate to the unknown switchpoint switching regression problem. When estimated, parameters generally conformed to theoretical expectations. The model was also tested for existence of switching regression behavior against a model similar to Goldfeld and Kane [12]. Significance exceeded 99% in all cases. With the advent of reserve intermediate targeting, it appears especially necessary to reinvestigate the behavior determining this important source of reserves.