ABSTRACTFatigue damage of low‐alloy steel samples tested earlier under uniaxial random loading spectra was evaluated using energy‐based models of Smith–Watson–Topper, Macha (M), Ellyin and Varvani‐Farahani with different descriptions in damage assessment. Damage over peak‐valley events of block loading histories was accumulated by means of these models. Smith–Watson–Topper approach involved stress and strain components on the maximum principal plane to evaluate fatigue life. M model related the life of samples to damage values calculated from the applied stress and strain histories. Ellyin model assessed damage of samples on the basis of dissipated hysteresis energy generated over fatigue cycles. Varvani‐Farahani damage approach assessed fatigue life on the basis of tensile and shear energies acting on critical plane over peak‐valley events of block histories. The predicted lives based on these approaches were compared with those of experimental data reported by M and coworkers. The choice of energy‐based models in damage assessment of steel samples was discussed on the basis of model description and terms of damage models.