The influence of muscle activation, position and velocities of body segments at touchdown and surface properties on impact forces during heel-toe running was investigated using a direct dynamics simulation technique. The runner was represented by a two-dimensional four- (rigid body) segment musculo-skeletal model. Incorporated into the muscle model were activation dynamics, force-length and force-velocity characteristics of seven major muscle groups of the lower extremities:​ mm. glutei, hamstrings, m. rectus femoris, mm. vasti, m. gastrocnemius, m. soleus and m. tibialis anterior. The vertical force-deformation characteristics of heel, shoe and ground were modeled by a non-linear visco-elastic element. The maximum of a typical simulated impact force was 1.6 times body weight. The influence of muscle activation was examined by generating muscle stimulation combinations which produce the same (experimentally determined) resultant joint moments at heelstrike. Simulated impact peak forces with these different combinations of muscle stimulation levels varied less than 10%. Without this restriction on initial joint moments, muscle activation had potentially a much larger effect on impact force. Impact peak force was to a great extent influenced by plantar flexion (85 N per degree of change in foot angle) and vertical velocity of the heel (212 N per 0.1 m s⁻¹ change in velocity) at touchdown. Initial knee flexion (68 N per degree of change in leg angle) also played a role in the absorption of impact. Increased surface stiffness resulted in higher impact peak forces (60 N mm⁻¹ decrease in deformation). It was concluded that changes in initial kinematic conditions are primarily responsible for variations in impact forces, and that muscular co-contraction does not play a major role.