Modern life-sustaining therapy often succeeds in postponing death but may be ineffective at restoring health. Decisions that influence the time and circumstances of an individual's death are now common and require an accurate and comprehensive characterization of likely outcome. Evaluation of alternative outcomes requires acknowledgement that most patients find some outcomes to be worse than death. Improved understanding of major predictors of patient outcome, combined with rapidly expanding technical abilities to collect and manipulate large amounts of detailed clinical data, have created a new intellectual and technical basis for estimating outcomes from intensive medical care. Such objective probability estimates, such as the system described here, can reduce uncertainty about difficult clinical decisions and can be used by physicians, patients, and society to reorient health care toward more scientifically and ethically defensible approaches.