This paper presents a new explanation for the post-announcement drift, i.e., the time-lag in market price adjustment to news surprises. Our explanation is that the post-announcement drift is a function of the private information agents have prior to news announcements. The more information people have about the "true" value of an asset – and the more they trade on this information – the smaller the abnormal return drift will be. We test this hypothesis empirically. Since private information is not observable by the econometrician, we use transaction-level data and a sequential trade microstructure model to calculate the probability of private information-base trading (PIBT) prior to an earnings announcement.