Cooperative information systems (CISs) are often characterized by a high degree of data replication; as an example, in an e-government scenario, the personal data of citizens are stored by almost all administrations. In such scenarios, organizations typically provide the same information with distinct quality levels and this enables providing users with data of the highest available quality. Furthermore, the comparison of data values might be used to enforce a general improvement of data quality in all organizations. In the DaQuinCIS project (Aguilera et al., 1999), we propose an architecture for the management of data quality in CISs; this architecture allows the diffusion of data and related quality and exploits data replication to improve the overall quality of cooperative data. In this paper, we present an overview of a component of the DaQuinCIS architecture, namely the quality notification service (QNS), which is used to inform interested users when changes in quality values occur within the CIS. QNS can be used to control the quality of critical data, e.g. to keep track of quality changes and to be always aware when quality degrades under a certain threshold. The interaction between the QNS and its users follows the publish/subscribe paradigm: a user willing to be notified for quality changes subscribes to the QNS by submitting the features of the events to be notified for, through a specific subscription language. When a change in quality occurs, an event is published by the QNS i.e., all the users which have a consistent subscription receive a notification. However, as shown in the paper by Marchetti et al. (2003), currently available pub/sub infrastructures do not allow to meet all the requirements that a QNS implementation should satisfy, in particular scaling to a large number of users and coping with platform heterogeneity. QNS addresses both these problems through a layered architecture that: (i) encapsulates the technological infrastructure specific of each organization; (ii) adopts the standard Web-service technology to implement inter-organization communications; and (iii) embeds solutions and algorithms (namely, merge subscriptions and diffusion trees) to reduce the use of physical and computational resources.- The remainder of this paper is organized as follows: we first introduce some preliminary concepts and the QNS specification; then we motivate and describe the internal architecture of the service. Due to the lack of space, explanations are given at a very high abstraction level. Interested readers can find technical and formal details, as well as running examples, in the paper by Scannapieco et al. (2003).