The invention belongs to the technical field of engineering design and data modeling, and discloses a degradation data-based product reliability assessment method and a parameter estimation method. The method comprises the following steps of stablishing a Wiener accelerated degradation model based on a random effect, wherein the drift parameter is a random variable obeying normal distribution; Constructing a residual life prediction model according to the accelerated degradation model, and deducing residual life distribution and a reliability function of the residual life; And according to theperformance degradation data in the stress state, through a two-step maximum likelihood estimation method, obtaining an unknown parameter estimation value of the model. According to the method, a newaccelerated degradation model and a new residual life prediction model are constructed through a classic Wiener model, and an estimated value of the residual life is obtained; According to the method, a two-step maximum likelihood estimation method is adopted, unknown parameters in the degradation model are estimated, restrictions of a traditional maximum likelihood estimation method are made up,optimal values of the unknown parameters can be obtained more accurately, the model can predict the remaining life of a product more accurately, and the application range is wide.