We investigate the different aspects of end-to-end multimedia services: content creation, server and service provider, network, and the end-user terminal. In the first part of the thesis we present the study of system level issues including standardization and interoperability, user interaction, and the design of a distributed video server. In the second part of the thesis we investigate the systems in the context of object-based multimedia services. We present a design for an object-based audio-visual terminal, some of the features of which have been adopted by the MPEG-4 Systems specification. We then present the study of the requirements for a file format to represent object-based audio-visual content and the design of one such format. The design introduces new concepts such as direct streaming that are essential for scalable servers. In the final part of the thesis we investigate the delivery of object-based multimedia presentations and give optimal algorithms for multiplex-scheduling of object-based audio-visual presentations. We show that the audio-visual object scheduling problem is NP-complete in the strong sense. The problem of scheduling audio-visual objects is similar to the problem of sequencing jobs on a single machine.We compare the problems and adapt job-sequencing results to audio-visual object scheduling. We give optimal algorithms for scheduling presentations under resource constraints. The constraints considered are the bandwidth (network constraints) and buffer (terminal constraints). We present algorithms that minimize the resources required for scheduling presentations. We investigate the structure of interactive audio-visual presentations by considering event specifications and event extents. We show that the only way to support interactivity is by reserving channel capacity to deliver the interactive components of presentations. We also present algorithms for computing the minimum auxiliary capacity required to support interactivity in objectbased audio-visual presentations.i