This study aimed to determine whether the time-course of maximal resting H-reflex amplitude (H(max)) recovery after a prolonged moderate-intensity muscle contraction differs according to the optimal stimulation intensity used (predetermined vs. readjusted). Thirteen males performed a sustained isometric plantar flexion at 40% of their maximal voluntary contraction torque output until exhaustion. H(max) of the soleus muscle was recorded before and 2, 6, 10, and 14 min after the end of the contraction, then normalized by the respective maximal M-wave to form the H(max)/M(max) ratio. During recovery, pre- and redetermined optimal stimulation intensities (mini-recruitment curve drawn before each recovery measurement) were applied randomly to measure H(max). When using redetermined stimulation intensities, normalized H-reflex values were systematically greater (+11, +16, +15, and +15% after 2-, 6-, 10-, and 14-min recovery periods, respectively) than those obtained with the predetermined intensity. Keeping the stimulation intensity constant to evoke H(max) after a sustained muscle contraction can underestimate the H-reflex facilitation occurring after exhaustive exercise. It is therefore more appropriate to redefine the optimal stimulation intensity to evoke H(max) (using mini-recruitment curves) when the purpose is to analyze spinal modulation during the recovery phase.