Background: Syncope is a transient loss of consciousness, also known as fainting. In contrast, presyncope is a sensation of fainting that does not result in loss of consciousness. Determining whether a person is likely to be affected by either of these is diagnostically challenging with the Head-Up Tilt Test (HUTT) and 24-hour Holter ECG. Not only is HUTT uncomfortable and somewhat invasive, and requiring special equipment, it also has limitations in accuracy for identifying neural mediated syncope (NMS). Similarly, traditional ECG characteristics may not indicate any pathology. This work is part of an ongoing effort to provide an alternative method, using heart rate variability (HRV) analysis Methods: We analyzed HRV over a 24-hour period in 28 participants with history of NMS and 47 with pre-syncope as classified by the treating physician. We applied wavelet analysis to the heart rate time series consisting of inter-beat intervals. Two wavelet types were used, Haar and Morlet, and real and imaginary parts were calculated, resulting in time series for each participant. These time series were summarised using 4 moments, notably the mean, variance, skewness and kurtosis. Results: We applied machine learning classification to these two groups based on the features of moments derived from the wavelet analysis. Kruskal Wallis tests were applied, resulting in significant differences between syncope and presyncope groups. These significant results persisted over a range of wavelet scales. Using the xgboost classifier, we were able to classify participants as either syncope or pre-syncope with an accuracy of 79%. Conclusion: Our results have implications for understanding these conditions and the possible links between disease and the autonomic nervous system. In addition, the high accuracy of the current method may provide an additional diagnostic tool, avoiding the expensive and invasive HUTT and enhancing patient comfort.