The use of natural and synthetic esters as transformer liquid insulation has increased significantly in the last 15 years. These insulating liquids have much higher flash points than mineral oils, reducing the fire risk of transformers. Moreover, esters are biodegradable materials and their use limits the environmental damages and risks in case of a leak into the ground or water. For this reason, synthetic esters are often used in transformers for electrical trains and offshore windmills. Despite the advantages offered by natural and synthetic esters, there are challenges for their generalized use, such as the scarce information available on the interpretation of the predictive maintenance tests. This paper provides a statistical analysis carried out on a database of oil analysis on railway transformers. Reference values are calculated for several maintenance tests as dissolved oil gas analysis or physical-chemical properties of the oil. One part of the tests corresponds to synthetic ester-filled transformers, while the other part is from mineral oil-filled units. All the transformers are installed in Spanish high-speed trains. The percentiles 90, 95 and 98 were calculated for the key gases, furanic compounds and physical-chemical properties of oils. The typical increase rates for the key gases were also obtained.