This work addresses the problem of fatigue strength prediction of crankshaft fillet rolling processes to improve its accuracy. It is empirical to usually consider the effect of fillet rolling process on crankshaft fatigue performance. The fatigue performance of rolling process is mainly determined by induced compressive residual stresses, increased hardness and reduced roughness. Because the first two factors are difficult to measure the arc surface of fillet rolled cranks, it is difficult to predict the enhanced rate of crankshaft rolled performance to baseline unrolled’s. In this work a prediction method of fatigue strength for ductile cast iron crankshafts rolling process is presented. This method indirectly predicts the effect of the increased hardness on fatigue performance by the resonant bending fatigue test and modelling of crankshaft fillet rolling dynamic for the induced compressive residual stress. The finite element (FE) model for the resonant bending fatigue test rig is validated by the dynamic stresses measured by the strain gauges. The dynamic model of rolling process is solved with an implicit finite element method (FEM) and validated by measurements of rolling displacements. Finally, this method is applied to a 1.5l I4 gasoline engine, in which crankshaft dynamic loads are calculated with Elastohydrodynamic (EHD) simulations.