STUDY OBJECTIVES: We hypothesized that continuous, automatic turning utilizing a patient-friendly, low air loss surface would reduce the incidence of early ICU pneumonia in selected groups of critically ill medical patients. DESIGN: Prospective, randomized, controlled clinical trial. SETTING: Medical ICU of a large community teaching hospital. PATIENTS: One hundred twenty-four critically ill new admissions to the medical ICU at Charity Hospital in New Orleans. INTERVENTIONS: Patients were prospectively randomized within one of five diagnosis-related groups (DRG)--sepsis (SEPSIS), obstructive airways disease (OAD), metabolic coma, drug overdose, and stroke--to either routine turning on a standard ICU bed or to continuous turning on an oscillating air-flotation bed for a total of five days. MEASUREMENTS AND RESULTS: Patients were monitored daily during the treatment period for the development of pneumonia. The incidence of pneumonia during the first five ICU days was 22 percent in patients randomized to the standard ICU bed vs 9 percent for the oscillating bed (p = 0.05). This treatment effect was greatest in the SEPSIS DRG (23 percent vs 3 percent, p = 0.04). Continuous automatic oscillation did not significantly change the number of days of required mechanical ventilation, ICU stay, hospital stay, or hospital mortality overall or within any of the DRGs. CONCLUSIONS: We conclude that air-supported automated turning during the first five ICU days reduces the incidence of early ICU pneumonia in selected DRGs; however, this form of automated turning does not reduce other measured clinical outcome parameters.