Objective In order to improve the sensitivity of the traditional FastICA algorithm to the initial weight, the proposed improved FastICA algorithm based on the over-relaxation factor was proposed to extract fetal ECG, in this paper. Methods Firstly, the maternal abdominal mixed signal in the Da ISy database was centered and whitened to remove the correlation between signals. Then, the over-relaxation factor was introduced into the Newton iterative algorithm to deal with the initial weights generated randomly, and then an improved FastICA method was proposed to extract fetal ECG. Finally, the results of fetal ECG extraction were evaluated by visual waveforms and quantitative indicators. Results The experimental results showed that the average number of iterations of the algorithm was reduced from 55 before the improvement to 15, the signal-tonoise ratio was improved, and the fetal ECG extracted from the improved algorithm was almost not doped with maternal ECG. Conclusion The proposed improved FastICA algorithm based on the over-relaxation factor, while maintaining the rate of convergence, relaxes the requirement of initial weight, avoids the unbalanced convergence, reduces the number of iterations. Therefore, the improved algorithm can extract a clearer fetal ECG.