Abstarct: In petroleum and groundwater applications field, reliable facies modeling is of paramount importance for exploration of new reservoirs and making right decision for management of resources. Many of the current cases of facies modeling are baxsed on variogram modelling. However, these conventional techniques, give the modellers limited capability to reproduce the continuities and geometries of facies modeling; and display more stochastic heterogeneity than expected. The reproduction of curvilinear and complicated patterns plays a crucial role in fluid analysis of the reservoir. Although object-baxsed methods have been used to gain this goal; they suffer from some limits, including hard data conditioning process and time and CPU demands. This thesis is a case study of Soroush oil field, located on the north of Persian Gulf, baxsed on facies modeling. Core and palynological data indicate that the Burgan B in the Soroush area consists of a series of incision-fill sequences occurring in an estuarine/coastal plain/ deltaic environment. Environmental analysis indicates that tidal and channel sediments have settled in this area. Multiple-point simulation of Burgan B in this field has been implemented utilizing SNESIM algorithm in SGEMS software. This algotithm accounts multi-point statistics and reproduces non-linear patterns, utilizing conceptual image. The Training Image contains the dominant conceptual patterns that geologists have presented in the domain. The Training Image utilized is a realization of object-baxsed techniques that has been produced by N.I.O.C Experts. This realization has been developed using well logs’ and core’s facies data; geology and sedimentology explanations. Results of Multi-point simulation model imply that patterns and continuity of available structures in the training image have been reproduced. This is also manifested in the information content and E-type maps. Although geometry of sinuosity channels of this field is reproduced, the numbers of specific patterns of the training image and the realization (Co-efficient of determination equals with 0/47) are considerably different. Co-efficient of determination between net to grass ratios and multi-point simulation realizations is 0.99. The respective figure for object-baxsed modeling equals 0.2. Reproduction of well samples in multi-point simulation are achieved with an average of 73% in the validation grid; and reproduction of nodes of the training image in the range of 70 to 80 percent in the middle laxyers and the average of 57 percent for other laxyers, which are acceptable. Oil- in-place in different realizations are within a small range (511-527 million barrels). By comparing the results of SIS and MPS, the uncertainty has been decreased by 5.5 times.