Benthic diatoms are widely used over the world to assess the quality of freshwater ecosystems. Their taxonomic diversity, life-history trait variety and ability to respond fastly to water quality degradation make them efficient ecological indicators. Even if diatom-based biotic indices are valuable tools in detecting the degradation/restoration of aquatic ecosystems, they do not clearly help to identify the nature of pressure(s) impairing in situ communities. In this context, we aimed at developing a diatom trait based tool that helps to estimate the probability that a specific pressure significantly impairs local communities. autecological information on various traits (biovolumes, life forms…) was gathered from literature and expert advices for species caught in > 1500 sampling sites of the French RCS national river survey. Most of this information was fuzzy-coded to take into account within-species variation in life-history characteristics. The responses of metrics were assessed for several pressure categories related to water chemistry (e.g. herbicides, acidification) and habitat degradation, in an ecoregional framework. For each pressure category a Condition Tree Forest model was built using the risk level as the response variable and the trait-based metrics as the predictive variables. This work contributes to the development of a multi biological-quality-element based diagnostic tool.