The construction of a facies model could be employed as a conditional data for any property simulation that results in a more reliable reservoir characterization in further steps. In this study, an Iranian gas reservoir with six wells was studied to determine the 3D reservoir facies model. Fifteen reservoir facies were first detected along one of the wells with detailed core and thin section descriptions. Due to the significant difference between the core and log data resolution, facies were clustered into four major groups regarding the digenetic processes and petrophysical lithofacies properties (permeability and porosity). The lithofacies specification effect on petrophysical properties distribution is usually the only criterion that is considered in conventional simulations. In this article, however, the diagenetic processes, which have a main influence on reservoir properties' variation especially on the permeability and porosity data, have been considered in reservoir facies clustering. After the clustering process, using the available logs and with the help of the neural network, facies distributions along the other five wells were also specified. To generate a 3D facies distribution into reservoir scale, sequential indicator simulation methodology was utilized with reference to the previously generated reservoir zonation from sequence stratigraphy, where acoustic impedance seismic attribute was set as the secondary data. This attribute was produced by the model-based inversion method applied in seismic cubes. Since the diagenetic processes have highly influenced the rock properties of this carbonate reservoir, their effects have been considered as the most important parameters in reservoir facies determination. The ultimate model properly honored each expected sequence cycle properties, such as diagenetic processes and lithofacies.