Abstract: Modelling of discrete fracture networks in naturally fractured reservoirs is a complex process that contains large amount of uncertainty. This study presents a novel methodology that integrates various features of geological, statistical and artificial intelligence techniques in a nested loop to characterize field fractures and then to model them. Properties of natural fractures such as size, orientation and aperture are of different scales and relevancies. Their characterization is, to some extent, technique dependent. Secondary properties such as fracture density and fractal dimension are therefore defined for better description of fractures� spatial distribution. Mathematical relationships between the primary and secondary fracture properties are non-linear and have not been fully understood. A neural network is incorporated in the proposed methodology to determine these relationships, by processing field data available from logs and core analyses. Combining the characterized fracture properties, a nested stochastic technique is used to simulate different degrees of heterogeneity and model discrete fracture networks in the whole reservoir. The dual application of neural network and fractal mathematics in discrete fracture modelling is innovative in this study. The result is expected to map more closely with the actual physical distribution of fractures and their properties than that are achieved so far with simplified stochastic-fractal approaches.