Humans are endowed with an ability to skillfully handle objects, like when holding a jar with the non-dominant hand whilst opening the lid with the dominant hand. Dynamic-dominance, a prevailing theory in handedness research, proposed that the non-dominant hand is specialized for postural stability, which would explain why right-handers hold the jar steady using the left hand. However, the underlying specialization of the non-dominant hand has only been tested unimanually, or in a bimanual task where the two hands had different functions. Using a dedicated dual robotic wrist interface, we could test the dynamic-dominance hypothesis in a bimanual task where both hands carry out the same function. We examined how left- and right-handed subjects held onto a vibrating virtual object using their wrists, which were physically coupled through the object. Muscular activity of the wrist flexors and extensors revealed a preference for cocontracting the dominant hand. Such stabilization action contradicts the dynamic-dominance hypothesis. While the reliance on the dominant hand was partially explained by its greater strength, the Edinburgh inventory was a better predictor of the difference in the cocontraction between the dominant and non-dominant hands. When provided with redundancy to stabilize the task, the dominant hand preferentially cocontracts to absorb perturbing forces.