We compared 560 adults hospitalized in our Medical Intensive Care Unit (MICU) to 245 ventilated babies hospitalized in our Neonatal ICU (NICU). Both ICUs had comparable mortality rates--roughly 1 patient in 5 died. The average length of hospitalization for nonsurvivors versus survivors was disproportionately short for NICU babies (13d v 33d) and long for MICU adults (15d v 12d). This phenomenon resulted in a redistribution of ICU bed-days and resources in favor of survivors for NICU babies (approximately 9 of every 10 NICU beds were devoted to babies who survived), and nonsurvivors for MICU adults (roughly 1 MICU bed in 2). Both ICUs had comparable percentages of patients predicted to die--roughly 1 patient in 3. The predictive power of an intuition of die was comparable--and not all that great. Almost one third of patients in both ICUs with a single prediction of "die in hospital" survived to be discharged. However, the likelihood of finding a neurologically normal NICU survivor after a prediction of "die" was only 5 in 100. To the extent that informed decisions can be made with 95% certainty, we may have found a foothold on the slippery ethical slope of benefit/burden calculations in the NICU. Unfortunately, we have no comparable data for MICU survivors.