The performance of empirical money demand equations over the past decade raises serious questions about money demand predictability. A variety of specifications were presented to explain past episodes of apparent money demand instability, but their success in predicting future money demand is limited in most instances. In particular, the unprecedented decline in the velocity of Ml during 1982 and 1983 was not captured fullyby any of the previously-modified conventional specifications. This paper evaluates a variety of the approaches and specifications proposed inprevious money demand studies to explain the behavior of the narrowly defined money stock from the mid 1970's through 1983. The empirical results cast doubt on the appropriateness of the conventional money demand specification in both the pre- and post- 1974 periods.