The assessment of the functional state of the autonomic nervous system (ANS) in real time, by means of spectral analysis of fetal heart rate variability, may serve to improve the diagnosis of pathologic conditions of importance to the perinatologist. The combination of two approaches, namely an efficient method for detecting fetal ECG from the abdominal maternal signal, followed by spectral analysis of heart rate variability, is tested as a new noninvasive tool to assess fetal viability in real time. This study demonstrates a pattern of ANS development via the spectral contents of heart rate variability. It is shown that during "quiet state," the "young" fetuses (gestational age = 23.5 +/- 1 wk) present twice as much power of heart rate fluctuations at all frequencies from 0.2 to 1.0 Hz as "mature" fetuses (gestational age = 39.75 +/- 1.5 wk). This finding is coherent with the evolution of a stable and mature ANS activity. At frequencies below 0.1 Hz, a 1/f alpha power law relationship (alpha = 0.85, r2 > 0.9) between spectral density and frequency is displayed in the two age groups. A respiratory peak has been observed in some of the short (64-s) traces we analyzed. However, no respiratory peak was ever observed in a long (256-s) trace, due to the episodic nature of the fetal breathing and immaturity of the ANS.