Abstract: To address the issues of neglecting joint probability characteristics of wind direction and insufficient fitting accuracy of traditional parametric wind field models in wind-induced fatigue assessment of standing seam metal roof systems (SSMRSs), this study proposes a full-process computational framework for SSMRS fatigue evaluation that incorporates nonparametric wind field statistical modeling. Taking a stadium project as a case study, the framework utilizes 45-year measured data from a meteorological station and employs Gaussian kernel density estimation (GS-KDE) combined with a Beta-kernel nonparametric Copula model to construct a high-precision joint probability density function (JPDF) of wind speed and direction. This approach achieves accurate fitting of local fluctuations in both wind speed and direction. On this basis, rigid-model pressure wind tunnel tests, full-scale static wind uplift resistance tests, and finite element simulations are integrated to establish a precise mapping from stochastic wind load input to structural stress response. Cumulative fatigue damage over a 50-year design service life is quantified using the rainflow counting method. Results indicate that when the wind speed-direction JPDF is considered, the maximum cumulative damage of the roof is 0.637. In contrast, the simplified method that adopts the most unfavorable wind direction yields a damage value of 3.784, which overestimates the fatigue damage by approximately six times compared with the JPDF-based result. It is confirmed that accurately incorporating the joint probability characteristics of wind direction is a prerequisite for reliable fatigue assessment of SSMRSs. The proposed full-process computational method effectively avoids the overestimation and distributional deviation of fatigue damage caused by traditional simplified approaches, and can serve as a reference for wind-induced fatigue evaluation of similar large-span roofing structures.