Abstract.This first study of Canadian securities' earnings forecasts published byInstitutional Brokers Estimate System(IBES) focuses on changes in the mean earnings per share forecasts of 159 to 188 companies from 1985 to 1987. Cumulative average residuals are used to detect the announcement effects of large earnings forecast revisions. The main results of this study are the following. First, an investor with access to changes of earnings per share forecasts at the beginning of the month of publication could realize abnormal excess returns. Second, trading strategies based on earnings forecasts revisions can also yield abnormal returns, but the magnitude of the revision, the sector of the company, and the month in which the revision is realized must be considered. Third, when financial analysts' forecasts are published, the informational content of large revisions in forecasts has already been discounted by the market. This result is similar to findings of U.S.– and U.K.–based studies. Finally, large forecasts revisions coincide with a period of abnormal returns. However, the information content of the announcement of forecasts changes cannot be established. The gains are larger if the trade is undertaken before the diffusion of the forecast revision to the IBES subscribers. These results do not vary with the model chosen to predict company returns. This does not necessarily indicate the existence of a market inefficiency because information acquisition and analysis costs, as well as transaction costs, may diminish considerably these abnormal trading ga