A fundamental challenge with solar energy is improving the reliability, and increasing the lifetime, of photovoltaic modules. Typically, photovoltaic module manufacturers guarantee 80% of the nominal power of their modules for 25 years, but this type of guarantee is not based on a deep comprehension of the real degradation mechanisms of the modules. Moreover, the current challenge is to extend module lifetimes up to 30-40 years. For these reasons, accurate predictions of the lifetime of modules are needed. The primary goal of this dissertation is to develop predictive models to evaluate the lifetime of photovoltaic modules, taking into account the climates in which they operate. We consider the technology of crystalline silicon (c-Si) modules, which largely dominate the field. In Chapter 1 we identify two main degradation mechanisms, namely, potential-induced degradation (PID), and interconnection failures, that are particularly of interest as they can occur non only in harsh environments such as desert or tropical regions, but also in temperate climates. The main topic of this dissertation is the PID failure mechanism. Chapter 3 offers an introduction to the topic of PID for conventional p-type c-Si modules. In Chapter 4 we develop a lifetime model for PID. Such model is based on empirical equations obtained from accelerated tests in laboratory. In particular, we introduce a dependency on voltage that allows to perform PID prediction at string level, and analyze in detail the regeneration mechanism under irradiance. The result is a set of equations that describe the main phases of the evolution of PID as a function of the stress parameters. In Chapter 5, this model is applied to predict the evolution of PID for devices operating outdoors, considering four locations with different climates. One difficulty in applying our model to outdoor conditions arises from the fact that stress levels vary continuously according to the meteorological conditions, while our model was developed from indoor testing at constant stress conditions. We overcome this issue by using a mathematical method based on the concept of equivalent time. Moreover, suitable thresholds on the weather conditions are set to properly simulate the different phases of PID. In Chapter 6, we employ accelerated stress tests to investigate which strategies (selection of materials and/or module architectures) would allow us to manufacture PID-free modules. Chapter 7 is devoted to the second degradation mode under study in this dissertation, namely, disconnection failures. We validate experimentally a model developed in LT-SPICE that simulates the performance of a string with disconnected ribbons. In summary, this thesis proposes a combination of accelerated test sequence and simulations that allow to predict, for a given location, the effect of PID on the module power output. We are confident that this methodology could be applied in general to other degradation mechanisms, thereby allowing an improvement of the prediction of photovoltaic modules reliability in different climate conditions.