Abstract Background Septic arthritis is a serious infection, but the results of blood and joint fluid cultures are often negative in children. We describe here the clinical features and management of culture-negative septic arthritis in children at our hospital and their outcomes. Methods We performed a retrospective review of a cohort of children with septic arthritis who were hospitalized at Children’s Hospital of Philadelphia between January 2002 and December 2014. Culture-negative septic arthritis was defined as a joint white blood cell count of >50000/μL with associated symptoms, a clinical diagnosis of septic arthritis, and a negative culture result. Children with pretreatment, an intensive case unit admission, Lyme arthritis, immunodeficiency, or surgical hardware were excluded. Treatment failure included a change in antibiotics, surgery, and/or reevaluation because of a lack of improvement/worsening. Results We identified 157 children with septic arthritis. The patients with concurrent osteomyelitis (n = 28) had higher inflammatory marker levels at presentation, had a longer duration of symptoms (median, 4.5 vs 3 days, respectively; P < .001), and more often had bacteremia (46.4% vs 6.2%, respectively; P < .001). Among children with septic arthritis without associated osteomyelitis, 69% (89 of 129) had negative culture results. These children had lower C-reactive protein levels (median, 4.0 vs 7.3 mg/dL, respectively; P = .001) and erythrocyte sedimentation rates (median, 39 vs 51 mm/hour, respectively; P = .01) at admission and less often had foot/ankle involvement (P = .02). Among the children with culture-negative septic arthritis, the inpatient treatment failure rate was 9.1%, and treatment failure was more common in boys than in girls (17.1% vs 3.8%, respectively; P = .03). We found no association between treatment failure and empiric antibiotics or patient age. No outpatient treatment failures occurred during the 6-month follow-up period, although 17% of the children discharged with a peripherally inserted central catheter line experienced complications, including 3 with bacteremia. Conclusions The majority of septic arthritis infections at our institution were culture negative. Among patients with culture-negative infection, empiric antibiotics failed for 9% and necessitated a change in therapy. More sensitive diagnostic testing should be implemented to elucidate the causes of culture-negative septic arthritis in children.