We used a particular computational approach, the uncontrolled manifold (UCM) hypothesis to investigate joint angle covariation patterns during whole-body actions performed by standing persons. We hypothesized that two kinematic synergies accounted for the leg/trunk joint covariation across cycles during a rhythmic whole-body motion to stabilize two performance variables, the trunk orientation in the external space and the horizontal position of the center of mass (COM). Subjects stood on a force plate and performed whole-body rhythmic movements for 45 seconds under visual feedback on one of the four variables, the position of the center of pressure (COP) or the angle in one of the three joints (ankle, knee, or hip). The Fitts-like paradigm was used with two target amplitudes and six indices of difficulty (ID) for each of the four variables. This was done to explore the robustness of kinematic postural synergies. A speed-accuracy trade-off was observed in all feedback conditions such that the movement time scaled with ID and the scaling differed between the two movement amplitudes. Principal component (PC) analysis showed the existence of a single PC in the joint space that accounted for over 95% of the joint angle variance. Analysis within the uncontrolled manifold hypothesis has shown that data distributions in the joint angle space were compatible with stabilization of both trunk orientation and COM location. We conclude that trunk orientation and the COM location are stabilized by co-varied changes of the major joint angles during whole-body movements. Despite the strong effects of movement amplitude and ID on performance, the structure of the joint variance showed only minor dependence on these task parameters. The two kinematic synergies (co-varied changes in the joint angles that stabilized the COM location and trunk orientation) have proven to be robust over a variety of tasks.