The aim of this study was to examine whether the indexes of heart rate variability (HRV) are stable from day-to-day in patients with vasovagal syncope and whether the stability of the HRV indexes is linked with that of the clinical results of the tilt test. Nineteen patients with a history of syncopal episodes and a positive tilt test underwent a second test 1 week later. Of these, 11 (group P-P) also had a positive second test, whereas 8 (group P-N) had a negative second test. Fifteen healthy volunteers served as a control group. Five time domain indexes were derived: the mean of all coupling intervals between normal beats (mean NN), the SD about the mean of all coupling intervals between normal beats (SDNN), the mean of all 5-minute standard deviations of NNs (SD), the proportion of adjacent normal RR intervals differing by >50 ms (pNN50), the root-mean-square of the difference between successive RRs (rMSSD) and the standard deviations of 5-minute mean NN intervals (SDANN). The control group showed good reproducibility of all HRV indexes (slope 0.86 to 0.97). The syncopal patients taken as a whole had significantly less reproducibility than the controls in the pNN50 parameter. This difference was due entirely to the patients in the P-N group, who had a remarkable lack of reproducibility in both the pNN50 and rMSSD measures (slope pNN50, 0.52; rMSSD, 0.78), whereas the P-P group had a reproducibility of all HRV indexes, which was no different from that in controls (slope 0.83 to 1.04). In patients with vasovagal syncope, certain HRV measures that express parasympathetic tone did not show the high reproducibility found in normal subjects. Syncopal patients who lack reproducibility in these HRV parameters also show a lack of reproducibility in the clinical result of tilt testing.