On the present work, we propose an algorithm that after unstable segments identification by the first derivative estimates the baseline of antepartum fetal cardiotachograms. Therefore, 50 records 5 minutes long, derived from abdominal ECG, were processed according to the methods of Dawes, Mantel, and our proposal. Besides visual analysis, statistical differences among methods were established More appropriate visual fits of the baseline were gotten by our proposal (76%) than Mantel (22%) and Dawes (2016). Since lower mean baseline was computed from Mantel (137.8+/-8.4 bpm) than from Dawes (140.4+/-8.4 bpm) and our proposal (140.0+/-8.4 bpm), different account for accelerations and decelerations were also found Our proposal showed advantages to estimate the baseline of cardiotachograms as it was properly adjusted at the beginning of the traces, depended of only 15 seconds of valid segments and corresponded more closely to the visual baseline.