Studies have shown that an inflammatory response may be elicited by mechanical ventilation used for recruitment or derecruitment of collapsed lung units or to overdistend alveolar regions, and that a lung-protective strategy may reduce this response. To test the hypothesis that mechanical ventilation induces a pulmonary and systemic cytokine response that can be minimized by limiting recruitment or derecruitment and overdistention. Randomized controlled trial in the intensive care units of 2 European hospitals from November 1995 to February 1998, with a 28-day follow-up. Forty-four patients (mean [SD] age, 50  years) with acute respiratory distress syndrome were enrolled, 7 of whom were withdrawn due to adverse events. After admission, volume-pressure curves were measured and bronchoalveolar lavage and blood samples were obtained. Patients were randomized to either the control group (n = 19): tidal volume to obtain normal values of arterial carbon dioxide tension (35-40 mm Hg) and positive end-expiratory pressure (PEEP) producing the greatest improvement in arterial oxygen saturation without worsening hemodynamics; or the lung-protective strategy group (n = 18): tidal volume and PEEP based on the volume-pressure curve. Measurements were repeated 24 to 30 and 36 to 40 hours after randomization. Pulmonary and systemic concentrations of inflammatory mediators approximately 36 hours after randomization. Physiological characteristics and cytokine concentrations were similar in both groups at randomization. There were significant differences (mean [SD]) between the control and lung-protective strategy groups in tidal volume (11.1 [1.3] vs 7.6 [1.1] mL/kg), end-inspiratory plateau pressures (31.0 [4.5] vs 24.6 [2.4] cm H2O), and PEEP (6.5 [1.7] vs 14.8 [2.7] cm H2O) (P<.001). Patients in the control group had an increase in bronchoalveolar lavage concentrations of interleukin (IL) 1beta, IL-6, and IL-1 receptor agonist and in both bronchoalveolar lavage and plasma concentrations of tumor necrosis factor (TNF) alpha, IL-6, and TNF-alpha, receptors over 36 hours (P<.05 for all). Patients in the lung-protective strategy group had a reduction in bronchoalveolar lavage concentrations of polymorphonuclear cells, TNF-alpha, IL-1beta, soluble TNF-alpha receptor 55, and IL-8, and in plasma and bronchoalveolar lavage concentrations of IL-6, soluble TNF-alpha receptor 75, and IL-1 receptor antagonist (P<.05). The concentration of the inflammatory mediators 36 hours after randomization was significantly lower in the lung-protective strategy group than in the control group (P<.05). Mechanical ventilation can induce a cytokine response that may be attenuated by a strategy to minimize overdistention and recruitment/derecruitment of the lung. Whether these physiological improvements are associated with improvements in clinical end points should be determined in future studies.