A voice coil motor apparatus applied to an auto-focus or variable-focus lens module of a miniaturized camera includes: a voice coil motor, a plurality of steel guidance shafts, a lens holder, and a base. The voice coil motor also includes a magnetic part including magnets and yokes, and an electrical part including a coil winding. The magnetic part is rigidly coupled to the lens holder to form the moving member. The electrical part and the steel guidance shafts are fixed on the base to form the non-moving member. The pre-loaded force due to the interaction between the leakage flux of the magnetic part and the steel guidance shafts eliminates the free-play resulting from the tolerance gap between the moving and the non-moving members, and that the magnets and the yokes form a non-circular magnetic structure having a rectangular contour in order to minimize the overall dimension.