The ambition of this paper is to analyse real exchange rate dynamics in Macedonia relying on a highly disaggregated dataset. We complement the indirect evidence reported in Loko and Tuladhar (2005) and we provide direct evidence on the irrelevance of the Balassa-Samuelson effect for overall inflation via service prices in the CPI. Furthermore, we estimate variants of the BEER model. We show that alternative econometric techniques and data definitions bear an impact on the robustness of the estimation results. Overall, productivity, government consumption and the openness variables were found to be fairly robust in terms of sign and size. An increase/decrease in the productivity variables is associated with an appreciation/depreciation of the real effective exchange rate. Given that the B-S effect admittedly has a very limited role to play through nontradable prices in the CPI, this relationship could be explained by the (inverse) quality effect proposed by Loko and Tuladhar and, possibly in addition to that, by the nontradable component of tradable prices.