Abstarct: Artificial intelligent system has wide range of applications in oil and gas industry. A key parameter in source rock characterization is Total Organic Carbon (TOC). This valuable parameter could be determined by Rock-Eval instrument that is time and cost consuming. By now, several methods including Artificial Neural Network (ANN), Fuzzy Logic, etc. are used for TOC estimation but all of them uses logging data to achieve the result. Seismic data covers comprehensive area rather than logging data in a well, so in this study by applying trained Neuro Fuzzy and integrating geochemical, logging, geology and seismic data, TOC section are estimated among whole seismic section. To achieve this goal, geochemical evaluation and thermal modelling was done in potential source rock formations in central part of Persian Gulf. Finally with consideration of all aspects such as availability, quality and quantity of data, Gadvan formation was selected for future steps. After evaluation of Rock-Eval data and thermal modelling in 6 wells of central part of Persian Gulf in Gadvan formation, it was appeared that this formation is not effective source rock. Maturity of Gadvan is in initial steps and will be increased through south of Block A. In the next step, TOC as a target log was estimated from logging data by ∆logR and Regression. The reason for using these 2 ways was insufficient data in Gadvan formation. Despite of insufficient data, the result was acceptable and correlation between estimated and actual TOC was good. In the next step, internal attributes were extracted from seismic data and correlation between these attributes and Acoustic Impedance (AI) as external attribute with TOC were determined. Finally 5 attributes including Amplitude weighted phase, Cosine instantaneous phase, Instantaneous frequency, Second derivative of Instantaneous amplitude and Acoustic impedance were selected for final input data in Neuro Fuzzy. Among these attributes, AI has very high correlation with TOC. Gadvan data including 5 attributes as input data and TOC as output data in all wells was trained in Neuro Fuzzy in the location of well and seismic tie. Training error was 0.0096 that is acceptable and correlation for testing data was 0.87 indicating good relation between predicted and actual TOC. Finally, by applying Neuro Fuzzy to seismic section in 2SK-1, SIA-1, 3H-1 and T-2, TOC section were estimated in Gadvan formation.