The article discusses problems of the empirical verification of the relative version of the theory of purchasing power parity based on aggregated price indexes (especially using the consumer price index). The goal of the articles is to compare empirical results obtained from cross-country time series analysis using cointegration analysis, Error Correction Model, as well as VAR analysis. The authors tested 21 currency pairs of the U.S.A., Canada, Japan, Switzerland, the Great Britain, Norway, and Sweden in the period between 1975 and 2007. All tested time series were cointegrated of the first order with the exception of the consumer price index of Switzerland, which was cointegrated of the second order. The results of cointegration analysis are not very robust. This is explained by the authors as follows: existence of high transaction costs in an arbitrage, existence of complementary goods and oligopoly structure of exports, and expected back reaction between exchange rate and inflation.