"Systematic patterns in returns following earnings announcements have been examined in a number of studies but have proven to be difficult to interpret. This study examines the earnings announcements for Hong Kong firms in an effort to provide additional insight into the phenomena of price reversal and drift by examining the effects of both the method used to identify winners and losers and also the length of the subsequent period analyzed. The two methods used to form portfolios are based on the magnitude of the formation attribute: earnings surprises or initial reaction. The observation period being analyzed extends from twenty trading days before the announcement date up to one hundred and twenty trading days subsequent to the event. Results show that both drift and reversal can be observed for the same sample and event. Like the US firms observed in other studies, drift is observed both in the short- run and the long-run for the Hong Kong firms. However, unlike other studies of US firms, reversal is only found in the long-run. More importantly, the appearance of drift observed is independent of the method used to form portfolios. Chances of drift and reversal occurrence are only ""slightly” higher when sample firms are grouped on the basis of earnings surprise. This finding presents a serious challenge to Ho, Liu, and Ziebart's (1998)contention that the portfolio formation method underlies the observed patterns."