Continuous monitoring of machinery and equipment is enhanced by the rise of Industry 4.0. With it, data acquisition becomes the main tool for fault prediction, supporting the incessant need for production and non-stop machine operation. Gears are pointed as a fundamental element of the operation, having their behaviour monitored for failure detection. These failures are usually related to punctual defects, which, although occurring after hours of operation, give off signs of their occurrence and evolution. The continuous monitoring aims at the early identification of these signals, avoiding the evolution of the failure until breakdown of the equipment is reached. This study presents the characterisation of the behaviour that contact fatigue failure of automotive gears induces in the gearbox vibration signal. The characterisation of the behaviour generated by the presence of the fault in the vibrational signal is fundamental for the development of continuous identification methods. To this end, this study performs damage induction on the flank of helical gears and then subjects them to testing. A controlled test centre is used to perform the tests, allowing control of the operation parameters. Three unidirectional accelerometers located in the gearbox are used to acquire the vibration of the system. Vibration analysis in the time and frequency domains are performed in order to characterise the behaviour that the failure presents in the vibrational signal. The study summarizes the main characteristics identified in the time and frequency domains besides highlighting the correlation between them. For this purpose, a vibration-based algorithm that correlates the previously mentioned characteristics is developed. Durability tests, promoting the natural evolution of this failure mode, were conducted in order to verify the capacity of the algorithm. This algorithm was applied in both tests, presenting a success rate above 80% correlating the vibrational characteristics with the occurrence of the contact fatigue failure.