Abstract: Remaining useful lifetime (RUL) estimation is a key issue in condition-based maintenance and prognostics and health management. In the literature about the RUL estimation based on degradation process modeling, the current methods only consider the cases in which the degradation path is linear or can be readily linearized. In this paper, a nonlinear degradation process modeling approach is proposed based on diffusion process and the RUL distribution is derived in the sense of the first passage time. It is observed that the proposed method can be used to describe a general nonlinear degradation path and the current linear approaches turn out to be its special cases. As for parameter inference, considering the fact that the true degradation cannot be observed directly due to measurement error, the measurement error is introduced into degradation modeling. Through the degradation measurement data, a maximum likelihood estimation method is presented to estimate the unknown parameters of the degradation model. Finally, we demonstrate the usefulness of the proposed model via laser data and gyro's drifting data. The results show that the proposed method can generate reasonably better results than the linear model and thus can be potentially applied in practice.