This paper shows the results of empirical testing of Balassa-Samuelson model functionality in case of Serbia and European Monetary Union. The research for Serbia pertains to the period between 2004 and 2010, while in the case of EMU, due to unavailability of data about employment per sector, the period is somewhat shorter, covering 2005 to 2010. Applying both methods of cointegration analysis (Johansen and Engle-Granger's one), we have reached the results which show that internal transmission mechanism does not function in Serbia.