Objective:To characterize hospital costs of pediatric intensive care and to determine which demographic and disease characteristics are associated with cost.Design:Prospective cohort study.Setting:A 20-bed pediatric intensive care unit (PICU) in an urban university-affiliated teaching children's hospital.Patients:All children (n = 1,376) admitted to the multidisciplinary PICU during the fiscal year 1994.Interventions:None.Measurements and Main Results:Demographics, diagnoses, organ failure, Pediatric Risk of Mortality score, length of stay (LOS), and outcome were recorded. All hospital charges were obtained. The actual hospital costs were calculated by two separate methods. First, we converted the itemized patient charges into costs, using corresponding cost-to-charge ratios for each charge. In addition, we examined all direct and indirect expenses for the PICU. Univariate and multivariate regression analyses were used to determine the correlates to cost. The study population was similar to that of other studies of pediatric intensive care. The PICU was 86% efficient. The total cost for PICU care was $16,983,323. Average cost per admission was $12,342 ± $22,313, and average cost per patient day was $2,264 ± $868. The cost because of the PICU location (room cost) was 52.1% of all costs, and cost of laboratory studies was 18.3%. Respiratory therapy, pharmacy services, and radiology each accounted for between 6% and 8%. Total cost was most closely related to LOS, but severity of illness (Pediatric Risk of Mortality), diagnosis, and organ failure were also significant. Severity of illness was the most important factor in determining the variation in daily costs. Increased severity of illness was associated with higher laboratory and diagnostic study costs. We found little difference in the PICU room cost when calculated by adding direct and indirect expenses, compared with that obtained by using the cost-to-charge ratio.Conclusions:The maintenance of the specialty location and its personnel is the most costly component of pediatric intensive care. The strongest correlate with total cost for pediatric intensive care is LOS, but if costs are normalized for LOS, severity of illness best explains cost variation among patients. These data may serve as the basis for additional studies of resource allocation and consumption in the future.