This research investigates the mean reverting behavior of international stock price indices using an errorcorrection approach. Estimations are conducted in both time series datasets of individual countries and panel datasets of many developed countries. The individual countries are: Denmark for the period 1922- 2010, Sweden for the period 1919-2010 and the United States for the largest period from 1871 to 2010. Two panel datasets are established – the pooled panel dataset of the above mentioned 3 countries for the period 1922-2010, and the panel dataset of 15 MSCI developed countries for the period 1971-2010. Constructing many possible proxies for the fundamental value of stock prices, we find strong evidence of mean reversion in stock prices for almost all proxies. The estimated speed of mean reversion varies across datasets, proxies and models but in general it is higher than all previous studies on the same topic. Applying the same approach of rolling-window estimation of Spierdijk et al. (2010), this thesis also confirms their findings on the dynamics of the mean reversion process. The highest speed of mean reversion is found during the time of World War I, the Great Depression and the start of World War II. These results imply that the speed of the mean reverting process of stock prices is dramatically higher in the periods of an unsustainable economy than those of normal economic conditions.