Grain size has an essential influence on the serviceability of metallic materials. In this paper, a noncontact laser ultrasonic testing platform is built to study the effect of copper grain size on the laser ultrasonic backscattered signal. According to the correlation between grain size and ultrasonic wavelength, the ultrasonic scattering by copper grains in the experiment contains not only Rayleigh scattering but also the transition region from Rayleigh scattering to stochastic scattering. Using time–frequency analysis, the influence of copper grain size on the characteristic parameters of backscattering was explored, and a prediction model of grain size was established, which was compared with the prediction model based on the attenuation method to verify the accuracy of the backscattering model. The results show that the backscattered signal can adequately characterize the grain size information and laser ultrasonics is a method that can realize on-line detection of grain size.