Commercial machines are now available to monitor the fetal electrocardiograph in labour (Cinventa, Sweden). We report our experience of the first one hundred women we monitored by this method. They were divided into five groups dictated by the change in the fetal scalp electrode used for monitoring and the changes made in the computer software used for signal processing. There were progressive significant improvements in the ability of the system to produce a continuous heart rate trace such that it became acceptable for routine fetal monitoring. The quality of the electrocardiogram improved to 50% of the 'check ECG complexes' being printed accurately. Electrical signal distortion causing baseline wander is however the significant remaining problem. On-line analysis of the T/QRS ratio improved very significantly from 36% to 84% of the monitoring time. There were no significant improvements in the recordings made in the second stage, which remained poorer in quality and reliability in all groups. As well as changes in the computer software, the use of a single helix electrode and practical experience contributed to the improvements.