A common finding in the literature is that systematic post-announcement drifts in security returns are associated with the sign or magnitude of unexpected earnings changes. This paper examines proposed explanations for these drifts. The paper also documents that the systematic drifts in security returns are found for only a subset of earnings expectations models. For a class of expectation models based on the time series of reported quarterly earnings, variables coding (i) the sign and magnitude of the earnings forecast error, and (ii) firm size independently explain 81% and 61% respectively of the variation in post-announcement drifts. The joint explanatory power of (i) and (ii) is 85%, indicating t h a t the effect, of these two variables is highly collinear. The drifts are a persistent phenomenon over the 1974 to 1981 period with no evidence of being concentrated in a specific subperiod. The properties of expectation models based on the time series of earnings are contrasted with earnings expectation models based on security returns. The latter exhibit no evidence of systematic post-announcement drift behavior. The expectation models based on security returns have the appealing property that the assignment of firms to unexpected earnings change portfolios better approximates the independence over time assumption. This property means that these models are less exposed to the “proxy effect” criticism that has been made of results previously reported in the literature. The results in the paper are based on a sample of over 56,000 observations covering the 1974 to 1981 time period.