The influence of rock mineralogy, climate, vegetation and relief on stream water composition and export as a result of weathering is a major scientific concern both at the local scale, because of the link between stream chemistry and biodiversity, and at the global scale, in relation to global climate changes. Environmental factors influencing stream chemistry at baseflow were studied for 95 forested catchments located in the granite part of the Vosges Mountains (NE France). Mean annual rainfall, slope, altitude, catchment area, forest cover and granite composition were attributed to each catchment using a geographic information system (GIS) and spatially distributed databases. These variables were then used to model stream water concentrations in SO4, Ca, Mg, K, Na, Si, ANC (Acid Neutralizing Capacity) and pH via linear regression analyses. Sulfate concentration is mainly negatively correlated to the mean altitude of the catchment. Silica and Na concentrations are strongly correlated and negatively related to the mean annual rainfall and not to altitude. pH is positively related to the sum of Ca, Mg and K, but not to SO4. For any type of granite, Ca, Mg and pH are generally negatively related to the mean annual rainfall, and positively related to the catchment area. Calcium is best explained by the rock content in hornblende, Mg by the rock contents of biotite and hornblende, K by the content in K-feldspar. Nevertheless models differ for each granite type, in relation to the availability of weatherable minerals and, probably, to water pathways. This study shows the interest of spatial analysis methods for understanding stream water concentrations at the mountain range scale in combination with local process studies and modelling at the catchment scale.