An in situ technique to determine the surface roughness of ultraprecision machining using optical characteristic effects is fundamental as probes are not used, which prevents contact damage on the surface. Because the plastic lens molding reprints the roughness of the mold core fabricated by machining, tool marks result in the poor surface of the plastic lens. The machined surface can reflect the input light of a green laser with a short wavelength of 532 nm. By varying the feed rate of the ultraprecision machining, several samples of the electrolyte-less nickel with different surface roughness have been examined using the green laser and the photodiode array. The distribution of the optical scattering effect of each sample under the light source of green laser is derived for prediction of the surface roughness. The results show that greater surface roughness produces more expansive distribution of light scattering. In addition, the bidirectional scatter distribution function (BSDF) of the machined surface is found to be proportional to roughness. Using the ratio of the main and the side measuring channels of the photodiode array, a suitable approach to establish the relationship between light scattering and surface roughness can be developed. The laser and the photodiode array are found to efficiently predict the surface roughness of the ultraprecision machined electroless nickel.