This dissertation examines whether the well-known under-reaction to quarterly earnings observed in stock prices (post-earnings-announcement drift) can be explained by analyst under-reaction. A value measure (P) is constructed based on long-term earnings forecasts from <math> <f> I/B/E/S</f> </math>. Returns are computed based on market prices and on P over a window starting from one quarter before the earnings announcement to four quarters after. The two return measures are compared at the portfolio level and at the level of individual firms. The results suggest that although both return series exhibit considerable under-reaction, analysts appear to under-react more than the market. It takes approximately two quarters for analysts to fully “catch-up” with the market. Given that analysts exhibit greater inefficiency than the market, I conclude that it is unlikely that the market drift could be explained by analysts' behavior. The new methodology developed in this paper helps to resolve conflicting results in prior literature. Additional sensitivity analyses suggest that the result is robust.