In some cases carotid sinus massage (CSM) may induce AF, whereas it may terminate AF in others. The purpose of this study was to investigate the influence of CSM on atrial fibrillatory frequency using spectral analysis of the surface ECG. Continuous ECG recordings were made in 19 patients (12 men, 7 women, mean age 61 +/- 11 years) with AF. Unilateral CSM was performed in the standard fashion to one randomized bifurcation of the carotid artery at a time. Ventricular rate and fibrillatory frequency were assessed in 30-second ECG segments at baseline and during CSM. The frequency content of the fibrillatory baseline was quantified using digital signal processing (filtering, subtraction of averaged QRST complexes, and Fourier transformation). CSM resulted in a relative change in fibrillatory frequency of 4.5 +/- 3.9% (range 0%-13%). In 8 (42%) patients an increase in fibrillatory frequency was found (6.4 +/- 0.5 vs 6.8 +/- 0.5 Hz, P = 0.012). In 9 (47%) patients a decrease in fibrillatory frequency occurred (6.5 +/- 0.8 vs 6.1 +/- 0.8 Hz, P = 0.008) without AF termination. The remaining two patients showed no change in fibrillatory frequency. CSM on the contralateral side after 2 minutes produced fibrillatory frequency changes in the same direction in all patients with a good reproducibility in its magnitude (r = 0.59, P = 0.05). Calcium channel blockers were more frequently used (78% vs 25%, P = 0.044) in patients with a decrease in fibrillatory frequency compared to patients with a frequency increase. There were no significant changes in ventricular rate during CSM. In conclusion, two different responses of atrial fibrillatory frequency to CSM were found. This might explain why CSM may facilitate AF induction in some cases and AF termination in others. Calcium channel blocker treatment may prevent an increase in fibrillatory frequency provoked by CSM suggesting a blunted electrical remodeling process.