Abstract: The unconventional tight reservoir has experienced multiple geological processes, the types of reservoir are varied, the shape and size are different, the distribution is discrete, and the lithology is complex, resulting that the conventional modeling can’t effectively solve the problem of the unconventional tight reservoir. Taking the characteristics of different scale pore, complex nature/ artificial fracture network in unconventional tight reservoir into consideration, this paper proposes a method that taking the characterization of micro-pore texture of unconventional tight sandstone and the techniques for identification and description of natural/ artificial fractures as basis, and contains the modeling of different scale pore and complex nature/ artificial fracture network. This method is mainly divided into 4 aspects, firstly, based on mercury injection data, the pore type reservoir is divided into 4 stages, and stochastic simulating the matrix reservoir distribution model controlled by sedimentary model; secondly, taking seismic data as basis, combined with advanced techniques such as ant-body, deterministic simulating the large scale natural discrete fractures networks controlled by human computer interaction; thirdly, according to the medium-small scale fractures by logging data and micro scale fractures by casting lamella data, calculating the fracture intensity curve, simulating the medium-small and micro scale discrete fracture networks by stochastic algorithm; finally, based on micro seismic data, simulating the different scale artificial fractures by combining the stochastic and deterministic algorithm. This paper take the upper sweet-point of development pilot area in Changji oilfield which is in Lucaogou Formation of Junggar Basin of Xinjiang as example, establishing the model of tight reservoir by the method above, highlight the characteristics of tight reservoir, providing the static model of multi-medium reservoir with high precision to numerical simulation.