Reservoir characterization is a process of describing various reservoir properties using all available data to produce reliable reservoir model to enable predicting reservoir performance. Reservoir characterization plays a crucial role in modern reservoir management. The most important reservoir properties include pore and grain size distributions, permeability, porosity, facies distribution, and depositional environment. The types of information needed for describing such properties are core analysis data, well log data, well test results, production and seismic survey data. Especially well log data can provide valuable but indirect information about mineralogy, texture, sedimentary structures and fluid content of a reservoir. Reservoir porosity and permeability are the two fundamental rock properties which relate to the amount of fluid contained in a reservoir and its ability to flow when fluid subjected to applied pressure gradients. These properties have a significant impact on oil fields productivity and reservoir management. However, porosity and permeability estimation from conventional well logs in heterogeneous reservoirs is a difficult and complex problem to solve by conventional statistical methods. Statistical methods have pitfalls in estimating these two parameters. Ahwaz oil field is an anticline with 67 kilometers length and 4-6 kilometers width that is located in southwestern of Iran. In this study, the provided information from 32 exploration wells was employed. Comparing results of estimating reservoir properties from well log data using different intelligent methods showed that Neuro-fuzzy technique has outperformed fuzzy logic and artificial neural networks (ANN) techniques. For all of these three techniques wer 32 wells data from Ahwaz oil field were employed and processed in MATLAB programming environment. The Inputed log data used in current study were caliper log (CAL), sonic log (DT), neutron log (NPHI) and density log (RHOB) plus effective porosity (PHIE) and water saturation (SW) estimates. In first step, a fuzzy inference system (FIS) was created using Sugeno method and Gaussian membership functions based on subtractive clustering. Back propagation algorithm was used for training, testing and checking data. The ANN system was feed-forward which trained by back propagation algorithm. Sixty, twenty and twenty percent of Data were used for network training, testing and validating respectively. The final predicted versus measured data analysis indicated that Neuro-Fuzzy has lower root mean squared error in predicting core porosity and permeability values compared to the ANN and Fuzzy logic employed in this study.