Ballet dancers have small soleus (SOL) H-reflex amplitudes, which may be related to frequent use of co-contraction of antagonistic ankle muscles. Indeed, SOL H-reflexes are depressed during co-contraction compared to plantarflexion at matched background EMG level. We investigated the effect of 30 min training of simultaneous activation of ankle dorsi- and plantarflexor muscles (co-contraction task) on the SOL H-reflex in 10 healthy volunteers. Measurements were taken during co-contraction. After training, there was a significant improvement in the ability of the subjects to perform a stable co-contraction. SOL H-reflex recruitment curves and H-max/M-max ratios were decreased after co-contraction training but not after 30 min of static dorsi or plantarflexion. The decreased H-reflex size correlated with improved motor performance. No changes in SOL and tibialis anterior (TA) EMG activity or EMG power were observed, suggesting that increased presynaptic inhibition of Ia afferents is a likely mechanism for H-reflex depression. In different sessions we measured SOL and TA motor evoked potentials (MEPs) by using transcranial magnetic stimulation (TMS), TMS-elicited suppression of SOL EMG and coherence between electroencephalographic (EEG) activity (Cz) and TA and SOL EMG. SOL and TA MEPs were depressed, while TMS-elicited suppression of SOL EMG and coherence were increased after training. Decreased excitability of corticospinal neurons due to increased intracortical inhibition seems a likely explanation of these observations. Our results indicate that the depression in H-reflex observed during co-contraction task can be trained and that repeated performance of tasks involving co-contraction may lead to prolonged changes in reflex and corticospinal excitability.