Background: Current compound prioritisation and monitoring approaches within Europe focus mainly on widely occurring priority and river basin specific pollutants but may overlook site-specific contamination from local emission sources. Thus, we propose a robust and semiautomated approach for the identification of site-specific chemicals and a prioritisation of water bodies with specific contamination based on non-target screening data from liquid chromatography coupled to high-resolution mass spectrometry. Results: For prioritisation of site-specific contaminants, we calculated rarity scores for all peaks occurring in a set of 31 surface water samples, which combine the maximum signal intensity of a peak in a dataset with its frequency of occurrence in that dataset in one single number. These were a robust measure without the need to address the problems of missing data in more sophisticated multivariate statistical methods. For our dataset, site-specific compounds were defined for rarity scores > 1000, and the studied 31 sites showed a huge difference in the number of such peaks (0-91 in positive and 0-48 in negative ion mode). Together with isotopologue detection, the evaluation of mass defects and the occurrence of homologue series, which all could be obtained from automated data processing, a more detailed characterisation of these site-specific contaminations was possible. For three selected sites with a high number of site-specific peaks, novel or unexpected compounds could be identified, which stem from specific usage or (former) industrial production upstream of these sites. Conclusions and outlook: The proposed approach allows for a rapid screening of large non-target screening datasets for site-specific contaminants, the prioritisation of sites with such a specific contamination and the subsequent identification of these compounds. Thus, the risk of overlooking possibly hazardous chemicals (including unknowns) which are not covered in conventional monitoring and prioritisation schemes is reduced.