A scoring system intended to assess mortality risk and permit surveillance, evaluation and comparison of medical care was developed in our Surgical Intensive Care Unit. Five simple clinical components were identified and assigned scores according to their statistically validated relationship to mortality and the summation of the component scores resulted in a daily System Outcome Score (SOS). Cluster analysis was used to divide the creation data set of 2,777 patients into suitable groupings of scores to predict mortality and the clustering was confirmed for reproducibility with a validation set of an additional 2,860 patients. Two patient care surveillance techniques were then developed. The first involved the definition of three unfavourable SOS patterns evolving during the course of a patient's admission and detection of one or more of these patterns permits identification of specific patients for whom review of care delivered may be appropriate. The second involves a global assessment of care utilizing the Outcome Index (OI) which relates overall mortality risk in the unit to the actual mortality rate over a given time period. The effectiveness of care can then be compared between different time periods within the one unit or between different units with similar patient mix. A simple menu driven program has been developed for the IBM personal computer and clones that facilitates data storage and retrieval, production of outcome reports and customization of the scoring ranges to meet local standards of performance.