A major conundrum in High-Throughput Screening stud-ies is the presence of frequent-hitters, which include non-selective com-pounds and molecules that are false positives in many screens. In ourstudy, we introduce a method to detect frequent-hitters specific to anassay technology using historical compounds’ structural information. Results from historic HTS campaigns, including artefact assays, arecurated on a cooperate database. Structural fingerprints are generatedfor each compound and used to train a machine-learning model able topredict the behavior of novel compounds.