Earlier studies have presented mixed evidence on the rationality of the Federal Reserve's preliminary money stock estimates. The authors investigate the rationality of M1A, M1, M2, and M3 for both seasonally and not seasonally adjusted data. They find preliminary growth rates of these aggregates to be rational for not seasonally adjusted data but irrational when data are seasonally adjusted. Using Monte Carlo studies, the authors conclude that irrationality in seasonally adjusted data arises from the specific seasonal adjustment procedure used by the Federal Reserve. As a result, researchers conducting similar tests may want to focus exclusively on not seasonally adjusted data.