An efficient technique for progressive lossless compression of volumetric data is described. It is based on the quadtree data structure and exploits the expected similarities between the neighboring slices. The proposed approach has better compression ratio than the octree method. A small amount of memory is required since only two slices need to be located in the memory at a time, which makes it suitable for hardware imple-mentation. A built-in compression in volumetric scanners could considerably reduce the transfer and storage requirements for volumetric medical data.