Data quality is a strategic matter in the context of e-government as the integration of services requires authentic, coherent, and reliable data. However, establishing databases that are devoid of duplication, redundancy, or ambiguity isn't simple either in theory or in practice. In the context of e-government, this problem has been neglected for too long, particularly because administrative databases have often been wrongly regarded as simple. We demonstrate in this chapter that this is not the case at all, in particular because of the questions of interpretation that they raise. This chapter is based on case studies stemming from the Belgian federal administration (social security, business directories, federal authentic sources, etc.). Contrary to the assertions of common theories postulating a permanent bijective relationship between data and the corresponding reality, we argue that an empirical information system evolves over time along with the interpretation of the values that it allows one to determine. To address data quality, we propose a temporal framework that provides new operational strategies to improve administrative data quality (mainly, new ways to define quality indicators for continuous monitoring and re-engineering strategies). We finally demonstrate how our approach is generally applicable in the context of empirical information systems. © 2011 Springer Science+Business Media, LLC. ; Chapitre 7. ; L'ensemble des travaux de l'auteur a donné lieu à la recension suivante publiée en 2011 par David Bade (Université de Chicago) :Bade David. It's about Time! Temporal Aspects of Metadata Management in the Work of Isabelle Boydens. In Cataloging & Classification Quarterly (The International Observer), volume 49, n° 4, 2011, pp. 328-338. ; SCOPUS: ch.b ; info:eu-repo/semantics/published