Background: Pediatric Index of Mortality 2 (PIM2) is a useful scoring system for the prediction of prognosis, and Oncological Pediatric Risk of Mortality (O-PRISM) for ICU support in children with HSCT. We investigated prognostic prediction and risk factors for survival through early detection of admission to ICU after HSCT. Methods: We reviewed retrospectively medical records of children cared for in ICU after HSCT between 2004 and 2010. Patients who died within 2 hours after admittance to ICU were excluded. We analyzed the worst parameters in ICU by a t-test, Cox-regression, multiple logistic regression and a receiver operating characteristics curve (ROC). Results: 54 children, with fifty-five disease events, were admitted to ICU after HSCT. Sixteen children were diagnosed with high-risk disease status and 8 with non-malignant diseases. Stem cells were sourced from 14 matched siblings, 22 unrelated donors and 19 cord bloods. The median duration in ICU was 8.5 days (0.5-110). The reasons for admission to ICU were 32 pulmonary, 14 neurologic, and 9 hemodynamic events. Six patients (11.1%) survived after intensive care. The factor indicating discharge with survival was mental status (p = 0.04), although other factors included FiO2, prothrombin time, potassium, and pupil reflex in univariate analysis. In multiple logistic regression, there were significant factors of PaCO2 (p = 0.028), O-PRISM (p = 0.039), and PIM2 (p = 0.004) for prognosis. For prediction of prognosis, O-PRISM (p = 0.019) was superior to PIM2 (p = 0.435) in intensive care children after HSCT. Conclusions: O-PRISM might be a predictable scoring system for children with ICU support, and the Glasgow coma scale and PaCO2 were more reliable prognostic factors in the post-HSCT period.