Cardiac neural control can be estimated by frequency domain characterization of RR interval variations. This technique may become a clinical tool, as autonomic dysfunction is involved in the pathophysiology of sudden cardiac death. The study was designed to investigate 24-h cardiac autonomic control in 104 healthy subjects aged 40-77 years and to evaluate the impact of gender, age, smoking and physical training level. The sympathovagal balance was evaluated by spectral analysis of RR interval oscillations. The square-root of power of the high- (HF; 0.15-0.40) and low-frequency (LF; 0.04-0.15 Hz) bands were considered indexes of the vagal function and of the sympathovagal interaction, respectively. In addition, the precise centre frequency of the LF and HF oscillations was determined. The vagal mediated respiratory-dependent HF oscillation exhibited a clear circadian variation, and obtained the highest power values during sleep. The centre frequency was significantly lower during sleep (0.26 Hz vs 0.28 Hz), probably due to a slower respiratory frequency at night. Values for vagal tone were higher in physically trained subjects throughout the 24-h, and decreased by 13% for every 10-year increase in age; it was also reduced in smokers. The amplitude of the LF oscillation exhibited no clear diurnal variation. The mean LF/HF ratio was 3.1. LF power was much higher in males, was reduced by 15% per 10 year increase in age, and was lower in sedentary and smoking subjects, throughout the 24-h. The mean centre frequency of this component was reduced with advancing age (0.08 to 0.06 Hz from age 40 to 80 years). LF and HF power correlated positively, (r = 0.68), and 46% and 16% of the interindividual variation in LF and HF power, respectively, was explained by the four factors (gender, age, smoking, physical activity). Thus gender, age smoking and physical training level have a significant impact on the power and centre frequency of the HF and LF components. These effects must be addressed in investigations on autonomic balance.