Due to a lack of an information history, IPO firms' information precision is not only generally low but also likely to be estimated initially with considerable error. I hypothesize and find that the deviation between expected and realized information precision is predictably associated with the magnitude and the persistence of long-run abnormal returns after an IPO. Specifically, an upward (downward) revision of information precision results in positive (negative) abnormal returns over the period in which investors update their beliefs. In addition, the positive abnormal returns of firms with unexpectedly high realized information precision are less persistent than the negative abnormal returns of firms with unexpectedly low realized information precision, which can extend up to 18 months after the IPO. The findings imply that long-term investors in IPO stocks do not necessarily behave irrationally, but that both positive and negative post-IPO abnormal performance is also consistent with rational investors gradually updating the perceived information precision parameter of these stocks.