Abstract Purpose Blood cultures (BCs) are key for pathogen detection in septic patients. We investigated the extent to which sampling was performed and what factors were associated with the absence of general or inadequate BC sampling. Methods We conducted a retrospective cohort study of hospitalized patients with sepsis admitted to one of three EDs in 2018. Primary outcome was the extent of general BC collection of at least 1 set. Secondary outcome was the extent of adequate BC sampling, defined as ≥ 2 sets before antibiotic therapy (AT). Multivariable logistic regression analysis was performed to identify factors associated with deficits in both outcomes. Results 1143 patients were analyzed. BCs were collected from 946 patients. Single BCs were taken from 520 patients, ≥ 2 sets from 426 patients. Overall, ≥ 2 BCs were taken from 349 patients before AT. BC sampling before AT occurred significantly more frequently when ≥ 2 BC sets were taken rather than a single one (81.9%, versus 68.4%, p < 0.001) and this also led to the highest pathogen detection rate in our cohort (65.6%). A body temperature of ≥ 38 °C was the a supporting factor for general and adequate BC collection in all three EDs. Retrospective analysis of 533 patients showed that the qSOFA score had no influence on general or adequate BC collection. Conclusion Data on everyday clinical practice in the pre-analytical phase of microbiological diagnostics shows considerable deficits and indicates the need for more implementation of best practice. The variations identified in BC sampling between EDs should be further investigated.