Artificial turf is being used more and more often. It is more available than natural turf for use, requires much less maintenance and new products are able to comply with sport performance and athletes' safety. The purpose of this paper is to compare the mechanical and biomechanical responses of two different artificial turf infills (styrene butadiene rubber, from granulated vehicle tires, and thermoplastic rubber granules) and to compare them to the performance of natural fields where amateurs play (beaten earth, substantially). Three mechanical parameters have been calculated from laboratory tests: energy storage, energy losses and surface traction coefficient; results have been correlated with peak accelerations recorded on an instrumented athlete, on the field. The natural ground proved to be stiffer (-15% penetration depth for a given load), and to have a lower dynamic traction coefficient (-48%); the different kinds of infill showed significantly different stiffnesses (varying by more than 23%) and damping behaviour (varying by more than 31%). In running, peak vertical accelerations were lowest in the artificial ground with thermoplastic rubber granules, while, in slalom, both artificial grounds produced higher horizontal peak accelerations compared to the natural ground. Results are discussed in terms of their implications for athletic performance and injury risk.