Modeling the time-behavior of critical performance variables or degradation levels plays an important role in the remaining useful life prediction and maintenance scheduling of real industrial equipment. Traditional approaches predominantly use the degradation data for the estimation of degradation model parameters, but lose sight of potential information of failure events. In this paper, a comprehensive assessment about the quality of the parameter estimators in terms of the nature of available information is given by leveraging the Wiener process and the gamma process. Under consistent algorithmic logics, the estimation strategy is amply discussed for three data patterns, that is, pure degradation data, pure failure data, as well as the combination of them. According to the inherent statistical characteristics and the concept of first passage time, the maximum likelihood estimates of all the unknown parameters are obtained either in analytical forms or in numerical coun- terparts. The precision of each estimator is assessed on the strength of several particular examples. As a quantitative indicator, the mean squared error of individual estimates is evaluated based on the Monte Carlo simulation. The implementation of this Monte Carlo simulation requires generating degradation and failure data that are associated with each other for every single system. To investigate the influence of sampling methods, three algorithms including random walk approximation, gamma bridge sampling and inverse transform sampling are adopted for the data generation. For the first two, it should be noted that the degradation trajectories are simulated via small virtual intervals rather than the actual interval between two successive inspections, with the purpose of seeking out relatively accurate failure times. Relying on the preset structures, the degradation path is truncated as soon as it exceeds the failure threshold. Di↵erent orders of magnitude of sampling interval are simultaneously tested for the horizontal comparison. For the last one, the samples of failure time are generated by employing the corresponding probability density function during the first step. After that, the mutually independent degradation increments are recursively simulated by referring to the specific conditional distribution. In general, the results preliminarily demonstrate that introducing proper failure information together with degradation information returned by regular inspections can contribute to the improvement of fitting ability and the risk reduction of model mismatch.