There is little doubt that moderate training improves cardiac vagal activity and thus has a cardioprotective effect against lethal arrhythmias. Our purpose was to learn whether a higher training load would further increase this beneficial effect. Cardiac autonomic control was inferred from heart rate variability (HRV) and analyzed in three groups of young subjects (24.5 ± 3.0 yr) with different training states in a period free of stressful stimuli or overload. HRV was analyzed in 5-min segments during slow wave sleep (SWS), a parasympathetic state which offers high electrocardiographic stationarity, and compared to data collected during quiet waking periods in the morning. Sleep parameters, fatigue and stress level checked by questionnaire were identical for all the three groups, with no signs of overtraining in the highly-trained participants. During SWS, a significant (p<0.05) increase in absolute and normalized vagal-related HRV indexes was observed in moderately-trained individuals compared to sedentary subjects, which did not persist in highly-trained athletes. During waking periods, most of the absolute HRV indexes indistinctly increased in moderately-trained individuals compared to controls (p<0.05), but did not in highly-trained athletes. Normalized spectral HRV indexes did not change significantly among the three groups. Heart rate (HR) was similar for moderately-trained and sedentary subjects, but significantly (p<0.05) lower in highly-trained athletes in both recording conditions. These results indicate that SWS discriminates the state of sympathovagal balance better than waking periods. Moderate training load is sufficient to increase vagal-related HRV indexes. However, in highly-trained individuals, despite lower HR, vagal-related HRV indexes return to sedentary values even in the absence of competition, fatigue or overload.