A service life prediction approach for roofing membranes and systems is proposed based on a probabilistic modeling of the time-dependent performance. The membrane performance is modeled using a discrete Markov chain that evaluates the change of performance over time as a result of deterioration or repair. This stochastic model will be developed using in-field performance data collected during roof inspections, considering the roofing system type, membrane type, age group, exposure conditions, and maintenance level. The use of in-field performance data captures the interaction between roofing components, as well as the synergism of the various degradation factors. The probabilistic modeling accounts for the uncertainty and variability of material properties, degradation factors, and quality of workmanship and maintenance. The service life is determined considering both the technical and economical performances, in addition to the risk of failure; this constitutes the first step towards the standardization of service life prediction of roofing membranes and systems.