A semi-mechanistic, semi-empirical erosion model was developed and validated by experimental data. The novelty of the developed model is by accurately predicting the impingement angle that produce maximum erosion and relating it to measurable mechanical properties such as hardness and fracture toughness. The ratio of the target material fracture toughness (R) to its hardness (H) is the parameter which governs the erosion mechanism and consequently the angle for maximum erosion. It was noticed that for material with high R/H, i.e. ductile materials, the preferred mode is ploughing, a process analogous to rubbing in which the penetrating particle squeeze the target material out of the indentation to preform ridges, which are knocked off at subsequent impacts. The lower the ratio R/H is, a tendency to fracture mechanism occurs. The fracture mechanism is a process very similar to brittle chipping during metal cutting, where the applied force propagates cracks parallel to the surface. A simplified model, defining the horizontal force which is applied on the particle in its movement through the material, was introduced. This simplified model takes into account the relative effect of each of the ploughing and fracture mechanism on the total material removal. The material parameter which governs the relative effect of the mechanisms is R/H. Particle angularity also contributes to decreasing of the fracture mechanism, hence a governing non-dimensional number was received, R/H.d(eff) where d(eff) is the effective diameter of the penetrating particle tip. Incorporating this parameter in the erosion model yielded good agreement with experimental data and was able to explain the phenomenon of different maximum erosion angle and velocity exponent for different target materials. The relation between the developed model. empirical coefficients to target material mechanical properties was studied for eight different materials, ranging from soft metals to hardened stainless steel. Correlations between model coefficients to R/H were developed and exhibited good agreement with mechanistic analysis. (C) 2016 Elsevier B.V. All rights reserved.