eservoir characterization is considered as the main primary step in defining three dimensional distribution of petrophysical properties needed for any reservoir modeling and simulation. These parameters are normally estimated using both laboratory scale tests carried out on core samples and well log data. Having more accurate and consistent petrophysical parameters used for modeling reservoir properties is vital for oil industry managers and hence it is highly demanded to have as much agreement with underlying actual subsurface geology as possible. Due to having more complicated relationship between porosity and permeability in fractured reservoirs, the estimation of theses parameters are more difficult and is still considered as challenging topic among upstream researchers. The reservoir under study is considered as fractured reservoir having low permeability in matrix part and highly permeable in crack paths. The total porosity and permeability is accounted for through addition of these two components normally called dual porosity system. Asmari formation as one of the main petroleum host in many Iranian south oil fields is an example of a formation with high effective porosity and permeability resulted from a naturally fractured network. The prime objective of this study was to evaluate the feasibility of using well log data as the predictor of estimating petropRhysical parameters of a fractured reservoir employing intelligent neural network and support vector regression techniques in Golshan gas field located 160 kms south east of Bushehr port. In the first step all required geological map and reports plus well log data and core sample results were obtained and put into a geodatabase ready for further processing. Then the data were corrected for any outlier and deviation followed by cross validating with corresponding geological data. The third step was to estimate formation water resistivity (Rw) and Fracture Intensity Index (FII) using available well log data. In the next step the predictor variables were selected as the inputs for estimating porosity and permeability along wells. In this study the available neural network and SVR codes in Matlab environment was modified to optimize the effective modeling parameters and improve their performances. The most important part of this study was to identify and estimate petrophysical parameters in fractured zones. The estimated aperture, permeability and porosity of cracks using corrected well log data showed that the probability distribution of these parameters follows the lognormal distribution having very rare data above 0.1 mm open aperture and 3 percent porosity in fractured zones. These fracture pattern and parameter values are in general agreement with known gas reservoirs documented in literature. The performance of neural network model in predicting FII along wells yielded 91.2 percent correlation between predicted and observed values in data training stage while this figure was 82.4 percent for test and validating stage. Results showed that if fracture parameters such as estimated open aperture, porosity and permeability values as extra inputs in neural network model, the network performance will be significantly improved. As the final step in estimating porosity and permeability parameters using well log data and other estimated fracture parameters the performance of both neural network and SVR results were improved reaching to 96.9 percent in training stage and 96.7 percent for validation and test data showing acceptable performance for both methods. Finally it is concluded that both neural network and SVR methods can be used effectively in predicting petrophysical parameters of fractured reservoirs such as Golshan one if the input predictor variables are selected properly