A previous study found that when humans hop on both legs with exoskeletal springs in parallel with the legs, net metabolic power decreases compared to normal hopping. Further, they retained near constant overall vertical stiffness. Here, I quantified the biomechanics and metabolic costs of 10 subjects (3F) who hopped on both legs normally and using a passive-elastic exoskeleton with three different spring stiffness profiles in parallel to the legs at 2.4-3.0 Hz. The springs had degressive (DG – stiff then compliant), linear (LN), or progressive (PG – compliant then stiff) stiffness. Compared to normal hopping (NH) at 2.4 – 3.0 Hz, use of the exoskeleton with DG stiffness reduced net metabolic power (P_met) by 13-24%, LN stiffness reduced P_met by 4-12%, and PG stiffness increased P_met by 0-8%. P_met was significantly reduced when using the exoskeleton with DG stiffness compared to NH at 2.4-2.6 Hz (p≤0.0135). Dimensionless vertical stiffness remained invariant while hopping with an exoskeleton compared to NH, except when using the exoskeleton with DG and LN spring stiffness at 2.8 Hz (p<0.005). Peak vertical ground reaction force was 9-24% lower (p≤0.0008) and center of mass displacement was 6-12% lower (p≤0.0013) at 2.4-3.0 Hz when using the exoskeleton with DG stiffness compared to NH. Hopping with an exoskeleton with DG stiffness provided the greatest elastic energy return (EE), followed by LN and PG (p<0.001). Future designs of passive-elastic exoskeletons used for bouncing gaits should consider using DG or LN stiffness profiles rather than PG stiffness to minimize metabolic costs.