Hemiparesis after stroke decreases ability to dorsiflex the more-affected ankle during walking. Increased strength would be beneficial, but the more-affected limb is often too weak to be trained. In neurologically intact participants, training one limb induces strength gains in the contralateral, untrained limb. This approach remains unexplored post-stroke. The aim of this study was to test the hypothesis that unilateral dorsiflexor high-intensity resistance training on the less-affected side increases strength and motor output bilaterally following stroke. 19 participants (84.1 ± 77.6 months post-infarct) performed 6 weeks of maximal isometric dorsiflexion training using the less-affected leg. Voluntary isometric strength (dorsiflexion torque, muscle activation), reciprocal inhibition (RI), walking ability (gait speed, kinematics, EMG patterns), and clinical function were measured within 1 week before and 4 days following training. Post-intervention, dorsiflexion torque increased by ~31 % (p < 0.05) in the more-affected (untrained) and by ~34 % (p < 0.05) in the less-affected (trained) legs. Muscle activation significantly increased bilaterally, by ~59 and ~20 % in the trained and untrained legs, respectively. Notably, 4 participants who were unable to generate functional dorsiflexion on the more-affected side before training could do so post-intervention. Significant correlations between muscle activation and size of RI were noted across muscle groups before and after training, and the relation between size of RI and level of muscle activation in the more-affected tibialis anterior muscle was significantly altered by training. Thus, significant gains in voluntary strength and muscle activation on the untrained, more-affected side after stroke can be invoked through training the opposite limb. We demonstrate residual plasticity existing many years post-stroke and suggest clinical application of the cross-education effect where training the more-affected limb is not initially possible.