Tool wear and tool failure are critical problems in the industrial manufacturing field since they affect the quality of the machined workpiece (unexpected surface finish or dimensional tolerance) and raise the production cost. Improving our knowledge of wear mechanisms and capabilities of wear prediction are therefore of great importance in machining. The three main wear modes usually identified at the tool/chip and the tool/workpiece interfaces are abrasion, adhesion and diffusion. Besides the fact that understanding mechanisms that govern these wear mechanisms are still incomplete, the experimental analysis is very difficult because friction interface features (such as temperature, pressure, particles embedded in the contact …) are not easily measurable. The objective of this research work is to understand the physical mechanisms governing the tool wear by taking into account the sensibilities to scale going from the microscopic scale (microstructure scale) to the macroscopic scale (scale of the manufacturing operation) passing by the mesoscopic scale (tool tip scale). For this purpose, an analytic wear model was developed to describe the abrasive wear and to predict the cutting tool life. The proposed model is based on a tribological approach including a statistical description of the distribution of particles seen as non-metallic inclusions. The latter are assumed embedded at the interface of contact and having a conical shape characterized by two main parameters in the present approach: the corresponding size and apex angle. The volume of the removed material per unit time is chosen in this study as the main parameter to describe the abrasive wear mode. Coupled with literature results, the developed model is able to take into account the nature of the sticking-sliding contact and the effect of the friction coefficient on the rake face of the cutting tool. In order to identify all the material's parameters of the predictive model, a study of non-metallic inclusion considered responsible of the abrasive wear was performed on the 42CD4 steel. The determination of inclusion type and inclusion morphology was assessed qualitatively and quantitatively through microscopic and MEB observations as well as image processing. Finally, the volume removed by abrasion given by the model was compared to the experimental results previously achieved under orthogonal cutting.