This report presents a model for relating readily available clinical and physiologic measurements to prognosis from mechanical ventilation. Using data from 571 acutely ill, ventilated patients admitted to the intensive care units of 12 hospitals, it illustrates the relationship between the disease, the initiating respiratory failure, the acute severity of the disease, and the patient's age and chronic health status and the patient's probability of survival. The results indicate that, of the 296 deaths among the 571 ventilated patients, 142 (48%) could be identified on the day of ICU admission to be at a 75% or greater risk of hospital death; 49 of these patients were estimated to have a 90% or greater risk of death. After 3 days of ICU treatment, estimates for hospital mortality increased to 97% (39 patients). We believe that such estimates, when available from a larger number of patients and combined with additional information on the patient's desires, expectations, preillness quality of life, and prognosis for long-term survival, can be helpful in decisions to withhold and withdraw mechanical ventilation.