Introduction: Pediatric pulmonary embolism (PE) is rare but associated with adverse outcomes. We aimed to characterize PE cases admitted in a tertiary hospital and to evaluate sensitivity of PE diagnostic prediction tools. Methods: Retrospective, descriptive study of PE cases admitted from 2008 to 2020. Data was collected from hospital records. Patients were grouped according to PE severity and setting (outpatients, inpatients). Associations with demographic characteristics, risk factors, clinical presentation, management and outcomes were analyzed. PE diagnostic prediction tools were applied. Results: 29 PE episodes occurred in 27 patients, 62.9% female, mean age 14.1 years. Most PE were central and massive or submassive. One was diagnosed in autopsy. Outpatients (n=20), admitted for classic PE symptoms, were adolescents; in half the diagnosis had been missed previously. Risk factors included contraceptives (65%), thrombophilia (35%), obesity (20%) and auto-immunity (20%). Inpatients´ PE (n=8), diagnosed during cardiorespiratory deterioration (n=5) or through incidental radiological findings (n=3), were younger and had immobilization (87.5%), complex chronic diseases (75%), infections (75%) and central venous catheter (62.5%) as risk factors. Retrospectively, D-dimer testing and adults’ scores performed better than pediatrics’ scores (sensitivity 92.9-96% vs 85.7- 92.9%). Both pediatrics’ scores missed a case with a positive family history. Discussion: Pediatric PE diagnosis is often delayed or missed. To improve it, the development of pediatric prediction tools as from validated adult scores merits to be explored. We propose that clinical presentation and risk factors may be different in inpatients and outpatients. Family history should be included.