Six subspecialists with considerable experience in fetal ultrasound viewed a selection of pre-recorded ultrasound scans. Scans from 18 patients recorded on VHS video-tape were supplied from five centres in the UK and Ireland, each made on a high-resolution ultrasound machine by an experienced sonologist at a referral centre. Each observer viewed the scans on a large display monitor in an individual viewing booth. The scans were viewed in random order, at randomly selected bandwidths. Observers, who were blinded to both recording and bandwidth, assessed the technical quality on a five-point Likert scale. They also recorded their diagnosis. The six observers each carried out 32 viewing sessions, which gave a total of 192 viewings. There was no significant difference in the perceived technical quality of the scans between the two bandwidths used (P = 0.09). Of the 84 recordings transmitted at 1920 kbit/s, 71 (85%) were diagnosed correctly or 'half correctly' and 13 (15%) were misdiagnosed. Of the 95 recordings transmitted at 384 kbit/s, 66 (69%) were diagnosed correctly or 'half correctly' and 29 (31%) were misdiagnosed. This difference was significant (P = 0.03). The results indicate that although there were no perceived differences in technical quality between recordings transmitted at 384 or 1920 kbit/s, diagnostic accuracy was marginally worse at the lower bandwidth. This suggests that the higher bandwidth conveys more detail and information to the observer, which in turn enables more accurate diagnosis. However, further work is required before a definitive choice can be made about the optimum transmission bandwidth for remote fetal ultrasound studies.