It is repeatedly noted that the existence of a stable demand for money function IS veryrncrucial for the conduct of monetary policy. Accordingly this study attempted to estimate arndemand for money illation for Ethiopia using quarterly data ranging from 1970171 :1 torn2003104:lV. The study tried to estimate both the long-run and short-run demand for realrnnarrow money balances using the Johansen multivariate framework by taking real GDP,rnthe saving deposit rate, the inflation rate and the depreciation rate of the birr in thernparaUel market as explanatory variables. The depreciation rate of the birr in the parallelrnmarket was, however, dropped from the long-run model as its inclusion could not result inrna meaningful co integration relationship. Thus, in the long-run demand for real narrowrnmoney holdings was found to depend positively on real GDP and negatively on the savingrndeposit rate and the one period-lagged inflation rate. The coefficient of real GDP was alsornclose to unity consistent with the quantity theory of money. Of the two opportunity costrnvariables in the model the inflation rate was found out to have a much pronounced impactrnon the demand for money reflecting the low level of development of the financial system.rnIn the short-run, only lagged values of money holdings and real GDP were found out tornhave significant impact on the demand for money. None of the opportunity cost variablesrnwere found to have any impact on the demand for money in the short-run. This is due to thernlow level of development of the financial system which limits the availability of manyrnalternatives to money holdings. This observation is corroborated by the very low coefficientrnof the error correction term in the vector error correction model which shows the speed ofrnadjustment to long-run equilibrium.