Center of mass perception was investigated by varying the shape, size and orientation of planar objects. Shape was manipulated to investigate symmetries as information. The number of reflective symmetry axes, the amount of rotational symmetry, and the presence of radial symmetry were varied. Orientation affected systematic errors. Judgments tended to undershoot the center of mass. Random error increased with size and decreased with symmetry. Size had no effect on random errors for maximally symmetric objects, but orientation did. The spatial distributions of judgments were elliptical. Distribution axes were found to align with the principle moments of inertia. Overall, judgments were very accurate.