Purpose <p>This study investigated the effects of running-induced fatigue on ankle plantar–dorsiflexion performance and explored the underlying neuromuscular control mechanisms.</p> Methods <p>Twenty-four male participants performed a right-sided ankle plantar–dorsiflexion task before and after running-induced fatigue, with simultaneous recordings of ankle peak torque, electroencephalography signals from the primary motor cortex (M1; Cz, C1, C2) and other brain regions, and electromyography signals from the tibialis anterior (TA) and medial gastrocnemius (MG). Ankle peak torque was used to assess plantar–dorsiflexion performance. Beta- and gamma-band corticocortical coherence (CCC) was used to evaluate brain network connectivity, while partial directed coherence (PDC) was applied to analyze the brain–muscle network.</p> Results <p>In the brain network, fatigue significantly reduced beta-band CCC for C1-Cz electrode pair (<i>p</i> = 0.022) during dorsiflexion, beta band for C2-Cz (<i>p</i> = 0.014), and gamma band for C1-Cz (<i>p</i> = 0.003) during plantarflexion. In the brain–muscle network, fatigue significantly decreased gamma-band PDC for TA-Cz (<i>p</i> = 0.007) during dorsiflexion and for MG-Cz (<i>p</i> = 0.010) during plantarflexion in the ascending pathway. The declines in ankle peak torques after fatigue were positively correlated with reduced gamma-band PDC in the ascending pathway (<i>p</i> &lt; 0.034).</p> Conclusion <p>These findings suggest that running-induced fatigue significantly reduced functional connectivity within bilateral M1 regions and the ascending pathway of the brain–muscle network during unilateral ankle plantar–dorsiflexion, with weakened connectivity in the ascending pathway linked to decreased ankle force output.</p>