We prove that 3-D profile mapping using white light interferometry has systematic errors caused by aberrations of the optimal system. The tilt of an object invokes an offset between the object and the reference ray, which eventually makes the aberration cancel incompletely. The fringe peak of a white light interferogram is mainly affected by the aberration effect while the envelope peak remains stable. By the difference between the two peaks, it is easily confirmed how much the error existed in the fringe peak. Experimental results prove that the error caused by aberration is varied by object tilt, microscope NA, optics alignment within the range of $pm$50nm.