To examine the clinical and pathologic features of acute hypoxemic respiratory failure in children. Retrospective review of medical records and pathologic material during a 44-month period. Multidisciplinary pediatric ICU. With the assistance of a computerized database, the medical records of 2,254 pediatric ICU admissions were evaluated to identify children with respiratory failure. Children with acute hypoxemic respiratory failure who met the following definition were selected for inclusion: a) PaO2 less than 75 torr (less than 10.0 kPa) with an FIO2 of greater than 0.5; b) diffuse bilateral infiltrates on chest radiograph; and c) exclusion of cardiogenic pulmonary edema clinically or by pulmonary artery catheterization. Patients were excluded if they did not receive tracheal intubation and assisted ventilation. The medical records were reviewed for demographic, clinical, and physiologic information. Pathologic findings from autopsy or lung biopsy were also reviewed. Patients were placed in one of six groups based on their underlying disorder. In addition, the presence of neutropenia, septic shock, or a history of bone marrow transplantation was noted as a coexisting condition. A total of 100 acute hypoxemic respiratory failure patients were identified (4.4% of all 2,254 pediatric ICU admissions; 50 male, 50 female). Mean age was 6.0 +/- 5.4 (SD) yrs (range 1 month to 18 yrs). The overall mortality rate was 72%. The mortality rate was not affected by the underlying disorder, but it was higher in the presence of septic shock (80% vs. 58%; odds ratio 2.8), neutropenia (88% vs. 64%; odds ratio 4.0), and bone marrow transplantation (95% vs. 66%; odds ratio 10.4). When multivariate regression analysis was performed using all coexisting conditions, human immunodeficiency virus status, and patient gender, only a history of bone marrow transplantation and gender appeared to affect outcome. Oxygenation ratio (PaO2/FIO2), alveolar-arterial oxygen tension difference, duration of exposure to high levels of oxygen, and airway pressure measurements indicated more severe derangement of pulmonary function in those patients who died. Cardiac function was similar in survivors and nonsurvivors. Respiratory failure occurred in 32 children with severe neutropenia (mean absolute neutrophil count 55 +/- 101 cells/mm3), including 16 children with an absolute neutrophil count of 0. Pulmonary tissue from 37 children was studied. Diffuse alveolar damage was observed in 24%; morphologic evidence of infectious pneumonitis was encountered in an additional 41%. Children with acute hypoxemic respiratory failure represent a heterogeneous subset of patients. In our group of patients, infectious pneumonitis was more commonly encountered than diffuse alveolar damage. The mortality rate of children with acute hypoxemic respiratory failure has not improved since 1980.