Most conventional statistical methods aiming at defining geochemical concentration thresholds for separating anomalies from background have limited effectiveness in areas with complex geological settings and variable lithology. In this paper, median+2MAD as a method of exploratory data analysis (EDA) and concentration-area (C-A) fractal model as two effective approaches in separation geochemical anomalies, are used to identify geochemical anomalies in the Takab 1:25000 geochemical data sheet by using stream sediment geochemical data and lithogeochemical samples. We compared the anomalous area created using thresholds from various methods on the same data against known deposits and lithogeochemical samples. Results indicated that EDA methods are more impressible by lithology and could not well identify the geochemical anomalies in the areas with variable lithology. On the other hand, the C-A model based on the distinct anisotropic scaling properties, was better in revealing local geochemical anomalies, because it considered the spatial characteristics of geochemical variables.