The authors retrospectively analyzed postransplantation events in 198 children who underwent hematopoietic stem cell transplantation (HSCT) between 1998 and 2002 to obtain a risk score for pediatric intensive care unit (PICU) admission and to ascertain variables predicting a poor outcome. Thirty-six patients (18%) were admitted to the PICU. Median age was 9 years (range 1-18). On univariate analysis, variables significantly associated with PICU admission were male gender (P = 0.01), more than first complete remission (P = 0.003), allogeneic transplantation (P = 0.001), engraftment syndrome (P = 0.03), and acute graft-versus-host disease grade of at least two (P = 0.05). According to this, patients were divided in two levels of risk (low and high), with a respective probability of PICU admission of 8.8 +/- 2.2% and 63.8 +/- 8.8% (P < 0.0001). Seventeen (47%) patients were discharged from the PICU. The probability of event-free survival after PICU admission at 3 years was 24.2 +/- 7%. On univariate analysis, variables with a negative impact on event-free survival were type of transplantation, inotropic support, a C-reactive protein level of at least 10 mg/dL, and a high O-PRISM score. On multivariate analysis, the only variable that influenced event-free survival was the O-PRISM score (< or =10 points, 54.6 +/- 15.3%; >10 points, 8.6 +/- 5.8%; P = 0.007). In conclusion, the risk of PICU admission may be easily estimated using simple variables. A high O-PRISM score at the time of PICU admission predicts a dismal outcome.