Prediction of the service lifetime of Photovoltaic (PV) modules is crucial for all PV stakeholders. PV manufacturers benefit from setting realistic warranties and researchers can improve the durability of PV materials and components in order to achieve a desired lifetime. To develop reliable predictive models, a thorough understanding of the different sources of uncertainties in degradation rates determination and lifetime prediction is indispensable. In this paper, we assess different sources of uncertainties in PV degradation rates and lifetime predictions using physical models. Three major sources of uncertainties are discussed: due to climate variables estimation, due to PV modules reliability and due to degradation rate models. Degradation rates evaluated using modelled and measured climatic variables are benchmarked to assess the uncertainties due to climatic input variables. It is found that module temperature estimation leads to the highest uncertainties in degradation rates of 1.7% to 14.5% (depending on the temperature model and also on the location) followed by relative humidity 2.4% to 12.2%. UV modelling showed the least effect on degradation rates variations of only 0.1% to 5%. The variations due to the different PV module reliability and degradation rate models are evaluated using measured PV performance data. It is shown that these variations can be as high as 65.5% and 59.4% respectively, even if the models are calibrated using the same data.