To ensure a structural design that combines inexpensive materials and efficient construction techniques with the highest safety requirements, knowledge of potentially occurring critical natural load cases is key. This applies in particular to buildings at risk, such as high-rise buildings or critical infrastructure. In this work, a special emphasis on wind-exposed structures is given, and the identification of natural wind processes and the resulting loads are analyzed. To overcome the dimensionality issue, an approximation procedure for high random dimensions based on the Probability Density Evolution Method (PDEM) is utilized to describe input multi-time Joint Probability Density Functions (JPDF) as well as evolutionary Probability Density Functions (PDF) of response quantities of interest. The basic idea is, to (1) have a detailed description of possible wind processes that can be translated to a load model, that is data-based, simulational, or hybrid. A detailed expression includes the multi-time JPDF. Then (2) a time-dependent reliability analysis of an exposed structure (e.g. a high-rise building or bridge) is performed. To achieve that specific threshold quantities (e.g. deflection limit) are predefined. Due to the available information in each time step, parts of the input JPDFs can be classified as safe because no critical threshold of the quantities of interest can be reached. For specific stochastic wind models, the reliability analysis is then already performed. In the last step (3) additional uncertainties regarding geometry, material parameters, or further loads (payload, traffic) are considered and a full direct probabilistic approach to estimate the reliability via PDEM is performed, here the goal is as well, to analyze the correlated influences of possible critical outcomes and estimate small failure probabilities.