To evaluate the efficacy of mirror biofeedback rehabilitation for synkinesis in severe acute facial paralysis in children. Eight pediatric patients with facial paralysis with an initial electroneurography (ENoG) value less than 10% who underwent mirror biofeedback rehabilitation (the child-rehabilitation group) were enrolled. Seven infants (under age 2 yr) who were unable to undergo rehabilitation (the infant-and-toddler control group) and adult patients (n = 13, range, 33-56 yr) who underwent rehabilitation (the adult-rehabilitation group) comprised the control groups. All the patients enrolled were baseline House-Brackmann (H-B) grade VI at onset. The patients began daily facial biofeedback rehabilitation using a mirror at the first sign of muscle contraction on the affected side and were instructed to keep their eyes symmetrically open using a mirror during mouth movements. The training was continued for 12 months after the onset of facial paralysis. The degree of oral-ocular synkinesis was evaluated by the degree of asymmetry in eye opening width during mouth movements. The synkinesis index was calculated as a percentage of the interpalpebral space width ([normal side - affected side]/normal side). Statistical analyses used non-parametric tests (the Kruskal-Wallis test and Steel-Dwass posthoc test). The synkinesis index was significantly lower in the child-rehabilitation group than in the infant-and-toddler control group or the adult-rehabilitation group (p < 0.001). Children who underwent mirror biofeedback rehabilitation had less synkinesis than the infant-and-toddler control group, suggesting that mirror biofeedback rehabilitation is more effective in preventing the exacerbation of synkinesis in children.