Recent years have seen a veritable boom in multimedia applications. Due to the ubiquitous computing paradigm and evolution, there is now a great need for distributed multimedia applications, in particular. However, the majority of these multimedia applications place high demand on quality of service (QoS) with respect to the communication and end systems. Until recently, most of these applications have been stand-alone applications: generally proprietary solutions which are either tailor-made for a specific communications systems or often extremely difficult to extend to other multimedia devices than those they are intended for. Hence, a prime motivation for this work lies in the recent giant steps taken in development of multimedia and communication systems and also in the advances made in designing and realizing distributed systems in general. It would be most effective if programmers were able to tackle the development of distributed multimedia applications using exactly the same, tried and tested methods as when developing conventional distributed applications, without having to worry about the data transfer between the various multimedia devices and the communications system or between multimedia devices. So, in this thesis, I argue that a distributed multimedia application can be designed by using the same principles as for any other distributed applications without sacrificing the need for quality of service. And, the aim of this thesis is to design and realize a multimedia device and stream management architecture which is based on distributed-object computing technology and, at the same time, ensures the quality of service required. This work was done in a field of application where the major concern was to support an organization's workforce in its life-long professional learning process through a mixture of traditional classroom-based face-to-face training, teletraining, and self-paced training programs such as CD ROM- or Web-based training. The requirements set for such applications are presented in this dissertation, because the applications themselves belong to a superior example of a distributed multimedia application. The first part of this work defines the notions of multimedia, communication systems, quality of service, and object-based distributed systems as the foundation of this work. Then, I go on to present an analysis of related work in research and standardization venues and identify a number of limitations in the work recorded in literature. The management architecture described in this dissertation is characterized by its openness, the fact that it may be efficiently integrated into heterogeneous systems, its scalability and, in particular, by its versatility. In order that the architecture might enjoy wide-spread deployment, we designed generic operating system-independent programming interfaces for the multimedia devices to be utilized and the quality-of-service-aware communications systems currently in use and implemented them in a prototype. A major contribution of this work is the design, implementation, and evaluation of the device and stream management layer of the architecture. Various multimedia devices and quality-of-service-aware communications systems were integrated into the architecture, and they demonstrate how simple it is when using the framework to rapidly and flexibly integrate new devices, which are then very easy to manage. All the different parts of the management architecture were designed, specified, and documented using UML as the formal modeling language. With the aid of an integrated tool, we were able to bring the techniques and components developed together in a prototypical implementation of the system, thus achieving concrete validation of the approach taken in this thesis. The research work discussed in this dissertation shows that applying the concepts of distributed application design to distributed multimedia applications by using distributed-object technology and by separating the transport stream from the control and management of the stream is a feasible option. It was proven both in the lab and in the production environment that the basic approach underlying this work is indeed viable and, more importantly, that even high-volume video streams could be handled without difficulty.