Motor imagery (MI) is defined as a mental simulation of movements without any actual motor action. In last decades, several studies had advanced our understanding of this mental process. They notably showed shared neuroanatomical substrates between MI and physical practice. These neuronal networks seem to be dependant of individual expertise. Although the scientific literature seems to agree for a reduced and focused brain activity as expertise increases, a few investigated complex whole-body movements because MI is usually studied through standardized common tasks (e.g., keyboard sequences). The present study aimed to explore cerebral activation during kinesthetic MI of an unpopular complex movement from the Parkour discipline, the “cat jump”, as a function of expertise (novices for whom the movement is completely unknown vs. parkour experts). We used a 256-channels electroencephalogram to measure cerebral activity during 60 trials of MI. We quantified MI time at the beginning and end of the experiment, and MI abilities every five trials. Preliminary behavioural results showed a shorter MI time as well as better and more stable MI abilities in experts than in novices. Preliminary neurophysiological data showed a more marked activity of the motor cortex for parkour experts compared to novices. The present study helps to understand and identify structural and functional changes due to expertise, and brings new insights onto a common question regarding MI training: how far do we need to know the movement to imagine it properly? These results participate to Brain Computer Interfaces development whose application are numerous (e.g., exoskeleton).