A very useful model for predicting abrasive wear is the linear wear law based on the Rabinowicz's equation. This equation assumes that the removed volume of the abraded material is inversely proportional to its hardness. This paper focuses on the stochastic modeling of the abrasive wear process, taking into account the experimental uncertainties in the identification process of the worn material hardness. The description of hardness is performed by means of the maximum entropy principle (MEP) using only the information available. Propagation of the uncertainties from the data to the volume of wear produced is analyzed. Moreover, comparisons and discussions with other probabilistic models for worn material hardness usually proposed in the literature are done.