The present study proposes a technological transfer from modern mining prospection to the field of archaeology, providing a methodology to facilitate the discovery of ancient mining sites. This method takes advantage of the thousands of geochemical analyses of streambed sediments, performed by national geological surveys to inventory mineral substances. In order to delineate geochemical anomalies, the datasets are treated following two different approaches: Exploratory Data Analysis and a fractal-based method often recognised as more powerful. Mineral prospectivity maps are then obtained by combining the results with a geographical information system. The surroundings of the Celtic oppidum of Bibracte, French Massif Central, known to have been mined at least since the Late Bronze Age until Modern Times, have been chosen to exemplify the method's potential in archaeology. First, an exhaustive record of the mining sites was undertaken over a pilot area by pedestrian prospection. If mineral prospectivity maps had been used as guidelines, ∼70% of these mines would have been discovered by prospecting only ∼15-20% of the whole area whatever the method used to treat the dataset. At least for our specific case, the multifractal approach is as powerful as EDA. Besides saving a significant amount of time and effort, the methods described here may supply clues for determining the nature of mineral substances exploited in the past, when such information cannot be straightforwardly obtained from the field or from textual archives. It should however be noticed that this approach is proposed as a first step before peer archaeological investigation following more conventional methods. Technically, there is no real obstacle to the application of the methodology proposed here, because (i) software and associated packages are freely available from the web, as well as original geochemical datasets (at least in France), and (ii) minimal mathematical skills are required.