Wear in rolling contact is a complex phenomenon because it is multi-scale and multi-physical by nature and involves many strongly correlated parameters. It makes the numerical calculation of wear very challenging. In order to simulate wear, a better understanding of the influencing parameters involved in the wear calculation is of importance. This work presents a new numerical calculation of railway wheel wear based on Archard's equation combined with a spatial statistic approach called the Universal Kriging technique. The influence of factors, such as the contact mechanics model and the wear coefficient were studied through numerical experiments. The outcomes provide a new insight into the roles of the contact modelling and of the uncertainties of the wear coefficient in the numerical calculation of wear under various contact conditions. By considering the uncertainties of the wear coefficient, the approach provides a min.-max. range for wear estimation instead of a single deterministic value; furthermore, it can provide a detailed wear distribution over the contact patch for damage analysis. In combination with an accurate model of wheel-rail contact, this new wear model offers a more realistic wear prediction compared to the methods presently used for wheel wear estimation in railway vehicles.