A user-controlled procedure was implemented for the structural analysis of geophysical maps. Local edge segments are first extracted using a suitable edge detector function, then linked into straight discontinuities and, finally, organised in complex boundary lines best delineating geophysical features. Final boundary lines may be attributed by a geologist to lithological contacts and/or structural geological features. Tests of some edge detectors, (i) horizontal gradient magnitude (HGM), (ii) various orders of the analytic signal ( A n), reduced to the pole or not, (iii) enhanced horizontal derivative (EHD), (iv) composite analytic signal (CAS), were performed on synthetic magnetic data (with and without noise). As a result of these comparisons, the horizontal gradient appears to remain the best operator for the analysis of magnetic data. Computation of gradients in the frequency domain, including filtering and upward continuation of noisy data, is well-suited to the extraction of magnetic gradients associated to deep sources, while space-domain smoothing and differentiation techniques is generally preferable in the case of shallow magnetic sources, or for gamma-ray spectrometry analysis. Algorithms for edge extraction, segment linking, and line following can be controlled by choosing adequate edge detector and processing parameters which allows adaptation to a desired scale of interpretation. Tests on synthetic and real case data demonstrate the adaptability of the procedure and its ability to produce basic layer for multi-data analysis. The method was applied to the interpretation of high-resolution airborne magnetic and gamma-ray spectrometry data collected in northern Namibia. It allowed the delineation of dyke networks concealed by superficial weathering and demonstrated the presence of lithological variations in alluvial flows. The output from the structural analysis procedure are compatible with standard GIS softwares and enable the geologist to (i) compare the structural features with various data sets in a common geo-referenced frame, and (ii) take advantage of attribute selection to highlight specific structural patterns or eliminate possible artefacts.