The acceptance and more widespread use of objective probability estimates will depend on addressing several practical issues. Physicians generally are unfamiliar with these estimates, their origin, and how they should be used. Physicians need to understand the conceptual origins of prognostic systems, their strengths and limitations, and their usefulness in helping to resolve uncertainty and improve critical decision making. Better communication and understanding of patient and family preferences and values will be necessary to ensure that the use of numerical predictions will not be the dominant or exclusive consideration. No matter how refined and elaborate these systems become, they represent simplification of even more complex biologic decision-making systems. The predictions must be interpreted carefully and used by trained intensivists, with the constant opportunity to exercise human discretion. The challenge to the developers of these systems will be to improve their accuracy, refine their use with specific disease entities that are important in determining outcome in critical care (e.g. multi-organ system failure, sepsis, adult respiratory distress syndrome), improve display of data and ease of data capture with automated systems, and construct data-bases and predictive equations that are fluid, constantly growing, and adjusting to capture developments in medical care.