The phenomenon of flying ballast is one of the most recent aerodynamic problems caused by the increase in train speed. This issue becomes extremely evident when the speed of the train is increased above 300 km/h. At such speed, the pressure and velocity fields of the flow generated by the train in the upper layer of the track could lift the ballast stones. Although it has been found that collisions begin to occur at speeds of 270 km/h or greater, there are no experimental evidence on the correlation between the speed and the actual number of impacts. In this work, an experimental methodology that is able to detect the number of impacts due to ballast lifting, by means of on-board microphones, is presented. It is shown how this method has been validated and then adopted to statistically analyse extensive experimental campaigns carried out with two types of trains and three different ballast conditions at speeds up to 360 km/h. Due to the copious statistical basis, the outcomes allowed to better correlate how the train speed affects the number of impacts that may occur during a run, to identify the regions of the train that are most subjected to the flying ballast and to both verify and quantify the effect of different ballast conditions.