Traditional geostatistical simulation methods assume that the first two order statistics are sufficient to model mineral deposits. However, these methods are not able to model some complex pattern present in geological environments and the spatial connectivity of high values. So, multiple-point simulation (MPS) methods were developed as an attempt to overcome these limitations. In this work a MPS approach based on wavelet analysis is applied to Olympic Dam deposit, to simulate both material types and grades. The method, termed wavesim, work as follows: first, it scans a training image with a template to generate a pattern database; then this database has its dimension reduced by applying wavelet analysis; after that, the patterns are divided into classes using k-means clustering algorithm, considering the approximate sub-band image; and finally the grid is simulated by comparing the conditioning data event in each node with the classes prototypes and choosing a pattern from that class. Olympic Dam's simulation results show that wavesim can be applied successfully to a large instance. The resulting simulated realizations are analysed and validated in terms of histograms, variograms and high-order statistics, the latter being performed by using high-order spatial cumulants (HOSC).