The efficiency and quality of abrasive machining processes influence costs and quality of elements produced, as well as whole products. Hence, the estimation of surface characteristics can greatly facilitate the requirements of performance evaluation. But the surface finish is dominated by many factors and physical model which could predict it is not practical. Therefore, it is very important to develop modelling techniques for the reliable prediction of surface characteristics.This thesis describes the development of a three-dimensional predictive modellling methodology for surface characteristics of a ground surface. The methodology uses solid modelers to generate the chips and remove them from the solid model of the workpiece. This results in surface model which represents the groud surface in three dimensions. Various surface characteristics can then be deducted from this model.The modelling of individual abrasive grains follows a statistical distribution which depends on the grinding tool characteristics. The cutting path depends on the relative motion between the tool and workpiece.The methodology was implemented for three different types of tool path. The results are in agreement with expected values.