In order to study the influence of distribution of geochemical elements on mineralization, this study uses Q-Q plot and cluster analysis to discuss the space distribution of elements in Liujiashan of northern Guangdong province. In contrast to traditional anomaly recognition method with iterative mean ±σ, this paper uses C-A fractal to separate anomalies from their background values, and calculates multi-fractal spectrum function. The C-A fractal analysis unites element content and frequency distribution of statistic method and singularity space configuration. It is useful for distinguishing chemical background and anomalous values which are relate to mineralization, and can provide significant clue for mineral potential evaluation.