Nowadays, maintenance of wooden window frames strongly depends on the practical experience of the parties involved. To get a better understanding of the required maintenance period and maintenance actions, a model has been developed to predict durability. This model makes it possible to optimize maintenance in relation to the required quality, taking into account the current value of degradation and factors causing degradation, which reflect the building variables and its surrounding. Since degradation processes have a strong stochastic nature, a probabilistic approach is applied in the model developed. This approach takes the stochastic nature of all relevant parameters into account and combines physicochemical relationships, test results, expert opinions and inspection results. The outcome of the model is a prediction of the degradation process in time, taking into account uncertainty. The model is incorporated in a web-tool with a focus on practicality (suggesting maintenance actions, compare effects of maintenance). In this paper the principles of this probabilistic model are described.