Structured light (SL) techniques are used for three-dimensional (3D) measurements of objects in a wide variety of applications. SL-based sensory systems obtain the 3D surface profile of an object from the deformation of light patterns projected onto the object of interest. The number of fringes used in the projected light patterns has a direct effect on the 3D reconstruction errors. This paper presents a novel methodology for determining the optimal sequence of multi-fringe patterns that minimizes reconstruction errors caused by random noise. Experiments conducted with a variety of objects as well as a comparison study demonstrate the effectiveness of the proposed methodology.