There has been extensive theoretical and empirical research on the subject of demand for money function. This particularly stems from the important role demand for money plays in macroeconomic analysis, especially in the design and implementation of monetary policy. The increase in studies, especially in developing countries, can also be attributed to a number of factors like: the impact of moving towards flexible exchange rate regimes, globalisation of financial markets, ongoing financial liberalisation, innovations in domestic financial products, the advancement in econometrics techniques and other country-specific events. This study estimates and examines the nature and stability of the demand for broad money (M2) in Botswana. This is particularly important in that the usefulness of a money demand function in the conduct of monetary policy depends crucially on its stability. The stability of the money demand function is crucial in that a stable money demand function would mean that the quantity of money is predictably related to a set of key economic variables linking money and the real economic sector. Therefore, this will help central banks to select appropriate monetary policy actions. Based on the findings, the study also proposes policy interventions. The vast majority of the literature on demand for money has underscored the fact that variable selection and representation, and the framework chosen are the two major issues relevant to modelling and estimation of the demand for money function. In modelling and estimating the demand for money function in Botswana, this study surveys a stream of theoretical and empirical literature on money demand in developed and developing countries, including countries that have similar financial sector similar to Botswana. Due consideration is also given to the macroeconomic and financial sector development in Botswana to help in the identification of the variables that are included in the demand for money equation. Most importantly, this helped in getting meaningful results that are free from theoretical and estimation problems. In particular, this study applied the multivariate cointegration approach as proposed by Johansen (1988) and Johansen and Juselius (1990) to estimate the relationship between broad money (M2), real income, interest rate, South African treasury bill rate, inflation rate and US dollar/pula bilateral exchange rate. The study obtains one unique long run relationship between money and the scale and opportunity cost variables. The coefficients of the long run relationship are then modelled along the general to specific approach as proposed by Campos, Ericsson and Hendry (2005). In this type of approach the general model is reduced by sequential elimination of statistically insignificant variables and checking the validity of the reductions at every stage to ensure congruence of the finally selected parsimonious model. In accordance with the economic quantity theory of money, the long run income elasticity obtained is 0.8021, which is close to the value one (unitary) suggested by economic theory. The coefficients of real income, exchange and inflation rate have the expected positive signs and were significant in the long run. Therefore, the long run demand for money (M2) in Botswana was found to be positively affected by real income, inflation rate and exchange rate. The lack of statistical significant of the own rate of money (88 day commercial bank deposit rate) and the foreign opportunity cost variable (South African Treasury bill rate) is attributed to multi-collinearity problems between these two interest rates. This could be caused by the fact that short term rates in Botswana are very responsive to movements in the money markets rates in South Africa. The short run dynamics of the demand for money function shows the slow speed of adjustment to equilibrium of about 2.9 percent in the first quarter and this is reflective of the lack of sufficient availability of banking services and the low returns on financial assets which could allow economic agents to re-establish equilibrium levels of money holdings faster. The final parsimonious model obtained clearly reflects a well specified stable demand for money function. Therefore, based on the findings we can be precise in stating that targeting a monetary aggregate can be a viable policy for the monetary authorities in Botswana.