Abstract It is well-known that internal defects play a key role in the Very-High-Cycle Fatigue (VHCF) response of metallic materials. VHCF failures generally nucleate from internal defects, whose size strongly affects the material strength and life. Therefore, S-N curves in the VHCF regime are defect size dependent and the scatter of fatigue data is significantly influenced by the statistical distribution of the defect size within the material. The present paper proposes an innovative approach for the statistical modeling of Probabilistic-S-N (P-S-N) curves in the VHCF regime. The proposed model considers conditional P-S-N curves that depend on a specific value of the initial defect size. From the statistical distribution of the initial defect size, marginal P-S-N curves are estimated and the effect of the risk-volume on the VHCF response is also modeled. Finally, the paper reports a numerical example that quantitatively illustrates the concepts of conditional and marginal P-S-N curves and that shows the effect of the risk-volume on the VHCF response.