Daniel L. Thornton briefly reviews the nature of the three standard dynamic models of money demand adjustment. The author points out that these specifications are not statistically comparable. Instead, he notes that these models should be compared in terms of their conformity with theory and their stability over time. Estimating these specifications for three subperiods between II/1951-II/1984, Thornton finds that (1) all three specifications are sensitive to the specification of the long-run demand for money, (2 ) none is consistent with theory over all three subperiods, and (3) none exhibits temporal stability. He concludes, however, that, since none of these specifications may adequately represent the short-run demand for money when estimated u sing aggregate data, the results may say little about the instability of money demand.