This paper uses a two-country model to show that the exchange rate pass-through (ERPT) coefficient must be used with reservation as a guide to macroeconomic policy. Not only its identification conditions are very restrictive (which increases the chances of assuming wrong hypotheses for its estimation) but it also cannot associate movements of exchange rates and import prices: depending on the origin of the shock an exchange rate devaluation is followed by an import prices increase or decrease. Furthermore, if the supply curve is subject to exogenous shocks the ERPT is not identified under the general assumptions of ordinary least squares (the most usual estimation procedure used in the literature). As a result, the ERPT is difficult to estimate and not much valuable as a rule of thumb for economic policy. However, this work shows the errors of using the ERPT as a rule of thumb are mitigated when its estimates are conditional on the source of the economic shock. Also, the investigation of alternative estimation techniques (using impulse-response matching, for instance) may solve the identification problem.