Analysis of heart rate variability has been used to study the effects of midazolam, morphine and clonidine on the autonomic nervous system, when administered to patients for premedication. Ninety-five patients were studied 60 min before and 60 min after premedication. Normal saline (n = 25), midazolam 0.08 mg.kg-1 (n = 24), morphine 0.15 mg.kg-1 (n = 23), or clonidine 2 micrograms.kg-1 (n = 23) were administered intramuscularly by random allocation. A Holter device was connected to the patient during the study period. Using power spectral analysis the low-frequency and high-frequency components were calculated from the Holter recordings. These are markers for sympathetic and parasympathetic activity respectively; the low- to high-frequency ratio was also calculated, a ratio of > 1 signifying sympathetic dominance. A significant reduction was noticed in both low-frequency and high-frequency power in the three premedicated groups, whereas no changes were observed in the normal saline group. In the case of midazolam, both the low and high frequencies were decreased but the low- to high-frequency ratio did not change significantly. Morphine and clonidine depressed the low-frequency component more than the high-frequency component and the low- to high-frequency ratio was decreased, suggesting parasympathetic dominance. We conclude that heart rate variability may be a useful tool for investigating the effect of drugs on the autonomic nervous system.