The use of progressive, on-demand image dissemination techniques can support efficient dissemination of very large images across networks. In this paper we examine the effectiveness of various design options in developing such on-demand dissemination systems. We show that the choice of the design options can have a profound impact on the efficient use of client, server, and network resources. Based on our performance evaluation experiments, we recommend that efficient dissemination can be achieved by tiling images with a session-based progressive wavelet compression algorithm, delivering the compressed data to tiles in a round-robin manner, and performing custom client-side virtual memory management for image data to improve image data retrieval speed.