In the forging industry, tools represent an important part in term of production and costs. Enhancing their life cycle is then a challenging issue. Several mechanical and thermal mechanisms are responsible for hot forging tools damage such as wear, thermal and mechanical fatigue. This work will be focused only on the mechanical fatigue life prediction for hot forging tools. Both experimental data analysis and numerical simulation will be discussed in this paper. The aim is to perform qualitative and quantitative indicators of mechanical fatigue. First, experimental data of fatigue tests are used to identify both plastic strain-based Manson Coffin and stress-based Basquin life laws for 2 tool steel grades. These laws are quite classical for fatigue prediction [1-4]. The half-life strain or stress amplitudes are usually used for their identification but these amplitudes are very expensive to obtain from a numerical point of view since it is well known that hot work martensitic steels present a continuous cyclic softening from the first cycle till the rupture. Therefore an important number of cycles have to be simulated to reach these mechanical parameters at half-life. For all theses reasons, an alternative methodology is used . The fatigue life curves are established using the mechanical parameters that are identified from the first hysteresis loops of fatigue experiments. Comparisons are performed with the fatigue laws coming from more classical identification procedure performed at half life cycle. Good agreement is shown between experimental data and the new laws. A lower scattering is even observed in experimental results in comparison to the traditional fatigue laws. Then these new laws are introduced in the commercial software Forge® and are then applied to different industrial cases. A pretty good agreement is observed between predicted tool life and industrial values. © Springer/ESAFORM 2009.