Abstarct: Conventional geochemical exploration models, which are used for determining anomalies in both local and regional scales, are generally baxsed on multi-variable statistical analysis, mextalometry, vertical geochemical zonality and other geochemical methods. In general, these methods have several limitations: 1) the separation of anomalies, which their data do not have normal distribution, is problematic, 2) the separation of anomaly from background, where the anomaly is associated with the formation mechanism, is difficult, 3) the results of various anomalies do not correlate on contour maps, and 4) the preparation of pattern recognition for various anomalies is not suitable. These limitations cause the process of interpretation to be time-consuming and costly. It seems that the method of neuro-fuzzy networks for specific mining geochemical exploration is very proper. Fuzzy theory considers uncertainties in recognition of multiple anomalies and artificial neural networks possess high capabilities in learning existing models. The combination of these two tools creates a suitable technique for recognizing multiple anomalies, and is capable of using standard exploration existing models and considering their uncertainties, and thus, reduces exploration costs and produces reliable results. This technique has been used in this thesis to discriminate dispersed molybdenum mineralization zones from blind molybdenum mineralization zones. The results obtained by applying this technique are in good agreement with the results obtained using conventional methods, and in addition, the technique does not contain the above limitations mentioned for conventional methods.