Rhythmic two-limb coordinated movements in the sagittal plane are variable and inaccurate when the movements are in the opposite direction as compared with those in the same direction (directional constraint). The magnitude of directional constraint depends on the particular limb combination. It is prominent in ipsilateral hand-foot coordination, but minimal in bimanual hand coordination. The reason for such differences remains unclear. In this study, we investigated the possible mechanisms underlying the production of the difference that depend on limb combination. Subjects performed two-limb rhythmic coordinated movements either in the same or in the opposite direction for three separate limb combinations (bilateral hands, contralateral hand and foot, and ipsilateral hand and foot). For each combination two different tasks were performed. In the first condition, subjects actively moved two limbs (active condition). Second, subjects actively moved one limb in coordination with a passively moved limb (passive condition). In the active condition, the directional constraint was dependent upon the limb combination, as reported in previous studies; the directional constraint was quite prominent in ipsilateral combinations, intermediate in contralateral combinations, and minimal for bilateral combination. However, differences in the directional constraint did not depend on limb combination for any combination in the passive conditions which apparently utilized closed-loop control. In other word, the difference depending on limb combination disappeared when control strategies become uniformly closed-loop. Thus, we speculate that the control strategy utilized depends on limb combination in the active condition. Additionally, different mechanisms other than closed-loop control also would have influence depending on the particular limb combination. This may result in differences in performance depending upon the limb combination.