OBJECTIVE: The study was designed to evaluate changes in autonomic nervous system function during induction of anesthesia with fentanyl, midazolam, and pancuronium and to answer the question of dose-dependency of these effects. DESIGN: Prospective, randomized. SETTING: A university hospital. PARTICIPANTS: Forty consecutive cardiac surgical patients. INTERVENTIONS: Anesthesia was induced with fentanyl, midazolam, and pancuronium. The patients were assigned to four groups differing in dosages of fentanyl plus midazolam and speed of injection. Fentanyl, 7.5 micrograms/kg (group A), 12.5 micrograms/kg (group B), and 20.0 micrograms/kg (group C) plus midazolam, 0.075 mg/kg (group A), 0.125 mg/kg (group B), and 0.200 mg/kg (group C) were administered over 10 minutes; in group D, fentanyl, 7.5 micrograms/kg, and midazolam, 0.075 mg/kg, were administered within 1 minute. MEASUREMENTS AND MAIN RESULTS: Heart rate variability (HRV) was measured using parameters in the time domain and the frequency domain. The comparison of preinduction HRV with the intra-anesthetic epochs did not show significant differences with respect to heart rate, coefficient of variation, and root mean squared successive differences. Spectral analysis showed significant reductions of power in the vasomotor band (0.01 to 0.05 Hz) and the low-frequency band (0.05 to 0.15 Hz) in all groups. Power in the high-frequency band (0.15 to 0.50 Hz) decreased slightly, but this did not reach the significance level. A dose dependency of these changes was found in the low-frequency band only. CONCLUSIONS: Parameters of HRV suggest that induction with fentanyl, midazolam, and pancuronium decreases sympathetic but not parasympathetic autonomic system activity. The anesthetic induction technique\'s modulation of autonomic nervous system balance is better represented by means of spectral analysis than by analysis in the time domain. This modulation was largely independent of the doses administered and independent of the speed of injection.