In process simulation, more reliable prediction of industrial results and evaluation of data for design purposes and process optimisation may be attained by the combined use of: the modelling and simulation of ore characteristics, and 2. the modelling and simulation of the process. It is well known that geostatistical methods allow the defining of models of orebodies in which the variables considered are the grades of the valuable components. These models are proper for reserve estimation, geostatistical contouring, design and simulation of the exploitation method, and, in addition, for evaluating the variability of the ROM which is expected to be fed to the processing plant. However, no data are normally incorporated in the orebody model in order to evaluate the ore characteristics relevant for a reliable process simulation. The procedure for process simulation aimed at predicting plant results normally used so far consists in collecting samples from different orebody sites and building up a so-called 'representative sample'. On this sample, the ore characteristics (eg liberation, washability, grindability, etc) and the processing tests are referred to as the 'average behaviour in the orebody', assumed constant. A more reliable methodology consists in collecting samples from different orebody sites, in performing tests to characterise the samples from the point of view of the process (ore patterns), and in using synthetic parameters (process features). The values of the parameters are processed using geostatistical methods in order to develop a model for simulating the feed to the process. The estimation may thus be referred to the amount of recoverable minerals instead of to the existing minerals. Furthermore, the selection of the processes, the sizing of equipment and the design of control devices, as well as the prediction of results, become more accurate and reliable.