An international, multicenter, prospective survey has been conducted to determine the hospital survival rates of patients with potentially reversible acute respiratory failure (ARF) who are managed in sophisticated intensive care units by leading critical care specialists, using current (1991 to 1992) support and treatment techniques and protocols. Twenty-five clinical centers participated in the survey, 11 in the United States and 14 in Europe. A total of 1,426 patients with ARF were studied, all of whom had been receiving closed system positive pressure mechanical ventilation at an FIo2 of at least 0.50 for at least 24 h at the time of entry into the survey. Of the survey patients, 793 (55.6%) survived their hospitalization, while 633 (44.4%) died in the hospital. The 1,426 patients with ARF were prospectively divided into two groups: group A, containing 375 patients, who at the time of entry into the study were hypoxemic or hypercarbic while receiving mechanical ventilator assistance; and group B, containing 1,051 patients, who at the time of entry into the study were neither hypoxemic nor hypercarbic while receiving mechanical ventilator support at an FIo2 of 0.50 or greater. Hospital survival rate for group A patients was 33.3% and for group B patients it was 63.6%. Survival rates were higher in patients with ARF caused by pneumonia (63%) or post shock lung injury (67%) and lower in patients with ARF caused by sepsis (46%). Severity of lung injury at the time of entry into the survey was a major prognostic factor, varying from an 18% hospital survival rate for patients with ARF with far advanced lung injury to a survival rate of 67% for patients with ARF with less severe lung injury. Low survival rates (<20%) were seen if mechanical ventilator FIo2 was 0.80 to 1.0, while 50% of the patients with ARF survived hospitalization whose FIo2 at entry was 0.50. Peak inspiratory pressure (PIP) >50 cm H2O at entry into the survey was associated with survival rate of less than 20%, while PIP <30 cm H2O was associated with survival rate of 60%. Shorter periods of mechanical ventilation (<48 h) of group A patients before entering the survey were associated with higher survival rates (38%) than patients requiring mechanical ventilation for more than 2 weeks (30%). Patients with ARF with multiorgan failure had lower survival rates (10%) than those with pulmonary dysfunction alone (45%). This survey indicates that management of patients in ARF requiring mechanical ventilator augmentation of blood gas transfer remains a serious challenge to intensivists today. High mortality rates are prevalent for patients with ARF, even in leading intensive care centers utilizing modern critical care techniques and protocols that are based primarily on high levels of mechanical ventilator support. Therefore, alternative methods for augmenting blood gas transfer that are now available and that avoid utilizing the already damaged and failing natural lungs and that allow less intensive mechanical ventilation should be assessed carefully and objectively. Utilization of such extrapulmonary blood gas transfer techniques may help improve the survival rate of patients with ARF.