A spring mass model is often used to describe human running, allowing to understand the concept of elastic energy storage and restitution. The stiffness of the spring is a key parameter and different methods have been developed to estimate both the vertical and the leg stiffness components. Nevertheless, the validity and the range of application of these models are still debated. The aim of the present study was to compare three methods (i. e., Temporal, Kinetic and Kinematic-Kinetic) of stiffness determination. Twenty-nine healthy participants equipped with reflective markers performed 5-min running bouts at four running speeds and eight inclines on an instrumented treadmill surrounded by a tri-dimensional motion camera system. The three methods provided valid results among the different speeds, but the reference method (i. e., Kinematic-Kinetic) provided higher vertical stiffness and lower leg stiffness than the two other methods (both p<0.001). On inclined terrain, the method using temporal parameters provided non valid outcomes and should not be used. Finally, this study highlights that both the assumption of symmetry between compression and decompression phases or the estimation of the vertical displacement and changes in leg length are the major sources of errors when comparing different speeds or different slopes.