We have examined the effect of isoflurane, halothane and fentanyl on heart rate variability (HRV) using power spectral analysis (PSA). Forty patients were allocated randomly to receive one of four anaesthetic techniques. Anaesthesia was induced with propofol 2 mg kg-1 and all patients breathed a mixture of 66% nitrous oxide in oxygen. Twenty of these spontaneously breathing patients received 1.5% isoflurane and 20 received 0.75% halothane. Ten patients in each of these groups received fentanyl 1 microgram kg-1 before induction. PSA was performed on 5-min segments of beat-to-beat R-R interval data taken before and 5 min after induction. No significant difference was observed in time domain or spectral HRV measures between inhalation agents. Fentanyl was associated with significantly slower heart rates, although no difference was observed in the power spectrum of those receiving fentanyl. Analysis of pooled data from the 40 subjects showed that mean heart rate was unchanged and heart rate SD was decreased, as were total and individual powers in three frequency bands of the HRV spectrum (low 0.02-0.08 Hz, mid 0.08-0.15 Hz and high 0.15-0.45 Hz). A significant decrease in the ratio of high power to (low + mid) power indicated a shift towards sympatholysis with anaesthesia. However, a significantly greater depression of mid frequency components in comparison with low or high frequencies suggests that anaesthesia selectively depressed components of the HRV mechanism, independent of a general effect on autonomic tone.