||The aim of this thesis is to investigate the predictability of Central-European common stock returns. In the first chapter, using weekly data on the Czech, Hungarian and Polish major value-weighted indices in the period 1996:1 to 2002:12, it is shown that the index returns contain predictable components and that both the mean and volatility can be forecasted from the time-series of historical returns. In Chapter 2, the Johansen cointegration analysis is applied to the weekly data on the Czech, Hungarian, Polish and German equity market index price in the period 1998:1 to 2002:12. The results indicate that the four considered markets are cointegrated when prices are expressed in local currencies, whereas no cointegration was found for prices in term of Euro. In both cases, there is significant cross-country predictability, i.e. lagged returns from one market can be used to predict returns from at least one other market. The third chapter is dedicated to studying the impact of nonsynchronous trading on the predictability of stock returns. The Lo and MacKinlay (1990) econometric model of nonsynchronous trading is generalized to allow for an autocorrelated common factor. Finally, in Chapter 4 the economic significance of stock return predictability is evaluated. A dynamic trading strategy based on the maximally predictable portfolio is developed for monthly returns on Polish stocks in the time-period 2000:1 to 2002:12 and its performance evaluated using various market timing measures. The results imply economically significant predictability in the Polish stock returns.