Introduction: Prehospital sepsis alerts assist paramedics in identifying patients with sepsis and in communicating this diagnosis to receiving facilities. Following the prospective implementation study of our regional systemic inflammatory response syndrome-based alert criteria (Alert), the purpose of this sub-study was to determine the cause of Alert false negatives (patients without an Alert that subsequently met sepsis criteria in the Emergency Department (ED)). Additionally, the sensitivity of the Alert for detecting sepsis was compared to the Quick Sequential Organ Failure Assessment (qSOFA) and Hamilton Early Warning Score (HEWS). Methods: This study was an additional analysis of the prospective Alert implementation study. Included patients were ≥ 18 years old, transported by a regional Emergency Medical Service and met severe sepsis or septic shock criteria (SS/SS, 2012 Surviving Sepsis Guidelines) in regional EDs in 2013. False negative patients were identified prospectively and reviewed by comparing paramedic determined Alert status to the retrospective application of the Alert criteria to Paramedic Call Report (PCR) data. The Alert sensitivity was first calculated from prospective data, then retrospective sensitivities of the Alert, qSOFA and HEWS were calculated by retrospectively applying these tools to PCRs, using ED diagnosis of SS/SS as reference standard. Results: In 2013, 229 patients met SS/SS criteria in the ED and had PCRs available; 115 (50.2%) were male and median age [interquartile range] was 76.0 [63.0-84.0]. Of 229, 149 (65.0%) arrived in the ED without an Alert (false negatives) and 46 (30.9%) of these met Alert criteria retrospectively and were therefore missed by paramedics. Sensitivity of the Alert was 34.9% when applied by paramedics and 41.5% when applied retrospectively to PCRs. The retrospective sensitivities of the qSOFA and HEWS were 37.6% and 67.7%, respectively. Conclusion: In ED patients diagnosed with SS/SS who arrived with no Alert, the majority (69.1%) were missed by the Alert criteria, rather than by paramedic application of the tool. The Alert had a sensitivity of 34.9%. When applied retrospectively and compared to the Alert, qSOFA had similar sensitivity and HEWS had increased sensitivity. Future research should focus on deriving improved alerts or implementing those with higher accuracy, such as HEWS.