The paper investigates the mean-reverting components in real stock prices for 16 countries. The temporary and permanent components of real stock prices are identified through appropriate restrictions on a vector autoregression of real stock returns and inflation. The multivariate time series technique identifies the size and significance of the mean-reverting component. The evidence supports the mean-reversion hypothesis that stock prices are not random walks. A significant temporary component in real stock prices of magnitude between 7 and 64% of the variation of quarterly real stock price movements is found. For a number of countries there is evidence of persistence in the temporary component.