Pediatric sepsis and bacterial infection cause significant morbidity and mortality worldwide, with immunocompromised patients being at particularly high risk of rapid deterioration and death. This study evaluated if PERSEVERE, PERSEVERE-II, or the PERSEVERE biomarkers, can reliably estimate the risk of clinical deterioration and 28-day mortality among immunocompromised pediatric patients. This is a single-center prospective cohort study conducted from July 2016 through September 2017 incorporating 400 episodes of suspected bacterial infection from the inpatient units at Cincinnati Children's Hospital Medical Center, a large, tertiary care children's hospital. The primary analysis assessed clinical deterioration within 72 hours of evaluation for infection. Secondarily, we assessed 28-day mortality. Clinical deterioration was seen in 15% of subjects. Twenty-eight day mortality was 5%, but significantly higher among critically ill patients. Neither PERSEVERE nor PERSEVERE-II performed well to predict clinical deterioration or 28-day mortality, thus we derived new stratification models using the PERSEVERE biomarkers with both high sensitivity and negative predictive value. In conclusion, we evaluated previously validated biomarker risk models in a novel population of largely non-critically ill immunocompromised pediatric patients, and attempted to stratify patients based on a new outcome metric, clinical deterioration. The new highly predictive models indicate common physiologic pathways to clinical deterioration or death from bacterial infection.