This paper uses recently developed econometric techniques to investigate the demand for money in Thailand. Initial estimates show the absence of any long-term, unique relationship among a monetary aggregate (M1 or M2), an income and a price variable. However, the introduction of an exchange rate in the above relationship shows the presence of a cointegrating relationship. This provides some support to McKinnon's hypothesis of currency substitution. It also raises some concern about the closed-economy focus of earlier studies on money demand in Thailand. Addition of a foreign interest variable to the above models provide some support to the capital mobility hypothesis. The long-run elasticity estimates for the M1 equation raise some doubts about the desirability of using this monetary aggregate as an intermediate target by the Bank of Thailand. The null hypothesis of a unit income elasticity and zero foreign interest elasticity are not rejected, but the hypothesis of price homogeneity is rejected. Moreover, the short-term forecasts of M1 are biased. On the other hand, the long-run elasticities for the M2 equation are within the expected range while the equation itself exhibits structural stability during the sample period and provides unbiased forecasts.