The recovery of infrastructure systems is of significant concern; in order to have effective risk management planning, an accurate prediction of the recovery time is required. A system may have different recovery paths due to the time of the accident, nature of the disruptive event, and surrounding environment, among many other factors. Hence, any model, which is employed to estimate the recovery time, should be able to quantify the effect of such influencing factors. Missing data, inappropriate assumption by analysts, and lack of applicable methodology are some practical challenges for recovery rate analysis. The purpose of this paper is to develop a methodology to address these challenges. It is based on the availability and the nature of historical data; it involves various steps, including categorizing the given data set into three groups: no or missing data set, homogeneous data set, and heterogeneous data set. Here, the Bayesian approach has been employed to handle the no or missing data set group. For the heterogeneous data set group, the proposed methodology suggested the application of covariate based models. Finally, for the homogeneous data set, the methodology employed statistical trend tests, to find the appropriate regression models. The application of the methodology is illustrated by real case studies.