Expanded public availability of U.S. banking data has prompted a need to reexamine end-of-quarter window dressing. We find substantial heterogeneity in the pattern of window dressing across banks and products, not all of which can be explained as customerinitiated, and some of which is consistent with theoretical predictions in the absence of publicly available data. These findings call into question the efficacy of financial disclosure in constraining banks?behavior, and raise new issues for further research.