Groundwater control is a significant issue during most underground constructions in a jointed rock mass. More than any other single factor, lack of groundwater control can be the major cause of extra cost and construction delays in underground construction. A proper estimate of the rate of inflow is critical in selecting tunnel alignment, determining the potential need for ground treatment (i.e. grouting) and/or lining installation, and obtaining an adequate schedule and cost estimate. In this thesis, a method has been adopted that is able to consider the exact fracture system of the field and thereby determine the water ingress amount into a tunnel. Hence, a code has been developed that takes advantage of FLUIUT 3D and FRACIUT 3D codes as base, which are designed to generate fracture networks and to solve flow equations in fracture systems, respectively. These base codes are capable of modeling fractures as circular discs and use pipe network flow model to solve the flow equations. The newly added code has the ability to consider the tunnel as a three dimensional cylinder which is a development comparing to past linear models in this media. As a case study, Long Zagros Tunnel was selected and its geological data is used to build the model. Window sampling was used to provide an area-based sample of discontinuities exposed at a given rock face. Four fracture sets including bedding was detected after the analysis of dip and dip direction values using stereographic projection. In order to find the distribution best fitted to measured discontinuity lengths, three frequently used probability distributions were evaluated from which lognormal distribution turne d out to be the best. Three dimensional fracture density, required as input parameter of the model, was obtained using a simple back analysis relating the desired parameter to the number of fractures intersecting the sampling window. Watson-Williams statistical test was conducted on multiple discrete fracture networks to choose those that their geometrical parameters are in accordance with the data collected from the field. A code is developed so that it is able to conduct Lugeon test, in order to calibrate hydraulic behavior of the model using the data obtained from field tests. Fracture aperture was the hydraulic parameter to be calibrated through this process. Permeability tensor of the field was determined and used to evaluate the uncertainty of the results of the model. Regarding the calculated permeability tensor, no preferential flow direction was observed and thus it revealed to be of no use in selecting the appropriate tunnel alignment. The comparison between acquired permeability and the Lugeon value used showed that empirical relations testify the results of the model better than analytical ones. Water inflow into the tunnel was calculated using the developed code in which an increase in amount was observed in comparison with the results of analytical methods.