Information in geological maps is not straightforward but often inferred based on concepts, definitions and intentions, making the assessment of its reliability challenging. In this thesis, we dedicate ourselves to the concept of uncertainty and ambiguity in geological maps, in order to develop concrete ways to deal with them. The pursuit of a "pre-modelling" strategy – uncertainty evaluation before starting to model the geological structure – resulted in the development of two methodologies. First, we propose a procedure to assess the general uncertainty of bedrock information in maps without going into the specificity of structural features. Secondly, we focus on the ambiguity in structural geological features, namely in the trace network. We propose a methodology – “Trace Information Extraction” (TIE) – that classifies traces so that their geometrical characteristics are more explicit and their ambiguity better confined. Both approaches run parallel with the goal of better managing the uncertain character of a geological map.