We performed visual comparison of 200 head magnetic resonance (MR) and 200 head computed tomography (CT) images compressed at two levels using standard Joint Photographic Experts Group (JPEG) irreversible compression and a preliminary version of the JPEG 2000 irreversible algorithm. Blinded evaluations by neuroradiologists compared original versus either JPEG or JPEG 2000. We found that this version of JPEG 2000 did not perform as well as the current JPEG for head CTs, but for MR images, JPEG 2000 performed as well or better. Around 7:1 compression ratio seemed to be a conservative point where there was no perceptible difference. Copyright (C) 2000 by W.B. Saunders Company.