When mammals run, the overall musculoskeletal system behaves as a single linear "leg spring". We used force platform and kinematic measurements to determine whether leg spring stiffness (k(leg)) is adjusted to accommodate changes in surface stiffness (ksurf) when humans hoop in place, a good experimental model for examining adjustments to k(leg) in bouncing gaits. We found that k(leg) was greatly increased to accommodate surfaces of lower stiffnesses. The series combination of k(leg) and ksurf [total stiffness (ktot)] was independent of ksurf at a given hopping frequency. For example, when humans hopped at a frequency of 2 Hz, they tripled their k(leg) on the least stiff surface (ksurf = 26.1 kN/m; k(leg) = 53.3 kN/m) compared with the most stiff surface (ksurf = 35,000 kN/m; k(leg) = 17.8 kN/m). Values for ktot were not significantly different on the least stiff surface (16.7 kN/m) and the most stiff surface (17.8 kN/m). Because of the k(leg) adjustment, many aspects of the hopping mechanics (e.g., ground-contact time and center of mass vertical displacement) remained remarkably similar despite a > 1,000-fold change in ksurf. This study provides insight into how k(leg) adjustments can allow similar locomotion mechanics on the variety of terrains encountered by runners in the natural world.