Phase-field approaches offer a unified framework to simulate fatigue-induced fracture{1, 2}, but practical adoption is often hindered by the wide variety of fatigue degradation functions and the lack of clear calibration guidelines. This contribution presents a systematic comparison of commonly used toughness-degradation-based fatigue phase-field formulations, focusing on Carrara-type threshold-controlled laws and Seleš-type energy-based laws within a unified implementation. To enable computationally feasible high-cycle simulations while retaining mean-load sensitivity, fatigue accumulation is performed in the cycle domain using an energetic envelope-load strategy. The framework is implemented with adaptive time stepping and adaptive mesh refinement to improve robustness and efficiency across long-life regimes and during rapid damage localization. A comprehensive parametric study quantifies how degradation function forms and internal parameters influence fatigue life predictions and the resulting S–N curve characteristics. Based on these insights, a practical calibration procedure is proposed that decouples slope control from life-level adjustment, improving parameter interpretability and reducing trial-and-error. The methodology is validated against experimental datasets for multiple metallic materials, including conventional AISI 1045 and additively manufactured 18Ni300 in both aged and as-built conditions, covering fatigue failure in dogbone specimens and crack growth behavior in compact tension specimens. The results show that different degradation functions can reproduce experimental trends, yet exhibit markedly different calibration flexibility and numerical behavior, leading to actionable guidance for selecting and calibrating degradation functions in engineering fatigue assessments.