One of the key objectives for mineral exploration is to map the most prospective areas for the selected deposit type at a regional scale, subsequently narrowing these down to camp size target areas and finally to individual prospects that can be developed into economic mines. Exploration targeting, at all scales, is constrained by conceptual mineral system models and requires the integration and interpretation of a variety of geophysical, geological and geochemical data. Geographic information system (GIS) applications are commonly used to integrate spatially referenced data sets to produce mineral prospectivity maps in support of the mineral exploration targeting process. A knowledge-driven fuzzy logic method is used in this work for data integration and prospectivity modelling for orogenic gold deposits in northern Finland. The modelling workflow is done in a stepwise manner at three scales. Each step simulates the successive stage used in mineral exploration from selecting the most prospective belt scale domain to the camp size target area and individual prospects. New, higher resolution data sets are added at each stage as is typically the case in mineral exploration. Northern Finland was selected as the test area due to the regional scale coverage of publicly available geoscientific data and its proven orogenic gold potential. The regional scale prospectivity map outlines the central part of the Paleoproterozoic Central Lapland Greenstone belt (CLGB) as the most prospective area for orogenic gold in northern Finland. The belt scale model of the CLGB improves the resolution mapping of several camp sized high prospectivity target areas. The camp scale model of one of these, a past producing mining camp, further narrows down the high prospectivity zones into prospect size targets. Three models have score Area Under Curve (AUC) values between 0.839 and 0.930 from the Receiver Operating Characteristic (ROC) method validation technique, indicating that the models are robust. The stepwise approach presented confirms that prospectivity mapping using the knowledge-driven fuzzy logic method is a scalable, flexible and relatively fast method which can be used for decision making at different stages of the exploration targeting process. Adding new data and rerunning the prospectivity model with the new data is fast and easy. The most important conclusion from the study is that remodelling individual regional and camp scale domains separately is highly recommended for mapping the most prospective target areas even if the data sets used are the same.