Limitations on the bandwidth to a server makes the retrieval of a large-size high-resolution image, even compressed, very time-consuming. To speed up this process we replace the retrieval process from a single server with the retrieval from multiple servers, i.e., simultaneously retrieving different part of an image from multiple server sites. In this paper, we present a block-based distributed retrieval strategy for images compressed by the Embedded Zerotree of Wavelet coefficients (EZW) algorithm. The experimental results validate that block-based distributed retrieval is an efficient strategy to speed up downloading.