This thesis arises from an engineering consulting engagement addressing the critical evaluation of fatigue strength estimation procedures in the very high cycle regime. It investigates the possibility that ISO 12107 based procedures may underestimate the B50 parameter, which represents the median of the fatigue strength distribution, thereby producing conservative assessments of component performance. To this end, the study examines different methodological approaches to the analysis of B50, clarifying the relationship between normative references, statistical assumptions, and the information content of the available experimental dataset. After an overview of the ISO 12107 framework, the estimation of B50 is carried out through parametric S-N modeling based on the Stromeyer formulation, expressed within a bivariate normal framework that captures stress and fatigue life variability.