Nonlinear ultrasonics is a powerful method for detecting microscopic damage and stress in materials. The method is based on introducing a perturbation signal at high frequency, and monitoring higher order harmonics for microscopic damage or quantifying the frequency shift for measuring stress. As the influence of microscopic damage and stress to the ultrasonic signal is weak, a highly precise measurement is needed. In this paper, errors in the measurement equipment, measurement methodology, and couplant type are experimentally quantified. Random errors and unknown systematic errors inherent to the method and its tools are defined. The measurement uncertainty in nonlinear ultrasonics is quantified. Based on the limitation of the measurement accuracy, minimum detectable stress level is defined.