The challenge of composing both sound and moving image within a coherent computer-mediated framework is addressed, and some of the aesthetic issues highlighted. A conceptual model for an audiovisual delivery system is proposed, and this model acts as a guide for detailed discussion of some illustrative examples of audiovisual composition. Options for types of score generated as graphical output of the system are outlined. The need for extensive algorithmic control of compositional decisions within an interactive framework is proposed. The combination of Tabula Vigilans Audio Interactive (TVAI), an algorithmic composition language for electroacoustic music and real-time image generation, with MIDAS, a multiprocessor audiovisual system platform, is shown to have the features desired for the conceptual outline given earlier, and examples are given of work achieved using these resources. It is shown that ultimately delivery of new work may be efficiently distributed via the World Wide Web, with composers' interactive scripts delivered remotely but rendered locally by means of a user's ‘rendering black box'.