The provisional molecular approach, proposed by EFSA in 2013, for the pathogenicity assessment of Shiga toxin‐producing Escherichia coli (STEC) has been reviewed. Analysis of the confirmed reported human STEC infections in the EU/EEA (2012–2017) demonstrated that isolates positive for any of the reported Shiga toxin (Stx) subtypes (and encoding stx gene subtypes) may be associated with severe illness (defined as bloody diarrhoea (BD), haemolytic uraemic syndrome (HUS) and/or hospitalisation). Although strains positive for stx2a gene showed the highest rates, strains with all other stx subtypes, or combinations thereof, were also associated with at least one human case with a severe clinical outcome. Serogroup cannot be used as a predictor of clinical outcome and the presence of the intimin gene (eae) is not essential for severe illness. These findings are supported by the published literature, a review of which suggested there was no single or combination of virulence markers associated exclusively with severe illness. Based on available evidence, it was concluded that all STEC strains are pathogenic in humans, capable of causing at least diarrhoea and that all STEC subtypes may be associated with severe illness. Source attribution analysis, based on ‘strong evidence’ outbreak data in the EU/EEA (2012–2017), suggests that ‘bovine meat and products thereof’, ‘milk and dairy products’, ‘tap water including well water’ and ‘vegetables, fruit and products thereof’ are the main sources of STEC infections in the EU/EEA, but a ranking between these categories cannot be made as the data are insufficient. Other food commodities are also potentially associated with STEC infections but rank lower. Data gaps are identified, and are primarily caused by the lack of harmonisation in sampling strategies, sampling methods, detection and characterisation methods, data collation and reporting within the EU.