Residential house price indexes (HPI) are used for a large variety of macroeconomic and microeconomic research and policy purposes, as well as for automated valuation models. As is well known, these indexes are subject to substantial revisions in the months following the initial release, both because transaction data can be slow to come in, and as a consequence of the repeat sales methodology, which interpolates the effect of sales over the entire period since the house last changed hands. We study the properties of the revisions to the CoreLogic House Price Index. This index is used both by researchers and in the Financial Accounts of the United States to compute the value of residential real estate. We show that the magnitude of revisions to this index can be significant: At the national level, the ratio of standard deviation of monthly revisions to the growth rate of the index, relative to the standard deviation of the growth rate in the index, is 29%, which is comparable to the relative ratio for other macroeconomic series. The revisions are also economically significant and impact measures used by policymakers: Revisions over the first 12 releases of the index reduce estimates of the fraction of borrowers nationwide with negative equity by 4.3%, corresponding to 423,000 households. Lastly, we find that revisions are ex-ante predictable: Both past revisions and past house price appreciation are negatively correlated with future revisions.