Knowledge of pediatric sepsis etiologies is needed to optimize empiric therapy. A retrospective cross-sectional review of 428 children with clinically diagnosed sepsis found that 13% had lobar pneumonia, 12% bacteremia and 10% viral infections. No etiologies were found in 76%. Empiric antibiotic coverage of vancomycin/piperacillin-tazobactam/gentamicin for immunocompromised children and vancomycin/nafcillin/cefotaxime for previously healthy children would have covered all bacteremic children.