The underlying mechanisms that lead to syncope are still unclear, especially in children. In this work, we applied a novel point-process model to study time-varying heartbeat dynamics and to characterize autonomic changes that occur prior to a syncopal event. Twenty-six children with history compatible with neurocardiogenic syncope (NCS) and a positive head up tilt table test (HUT) were included in the study. ECG and blood pressure signals were recorded during rest and the diagnostic HUT. Using self-developed software, a decrease of > 30% of the median systolic blood pressure during HUT compared to rest was selected as the onset of the syncopal event. After ECG peak detection and correction of ectopic beats, we modeled the time between R-wave events as a history dependent inverse Gaussian (IG) and applied the pointprocess framework to compute several measures related with HRV. We tested for significant changes in these measures for three consecutive two-minute time intervals previous to the syncopal event. Of all measures, only the mean of the heart rate probability density function, μ HR , and the scale parameter of the IG probability density function, ξ 0 (t), presented a statistically significant increase prior to syncope, providing novel features associated with the statistical properties of heartbeat generation that could be critical to predict and explain the occurrence of syncope.