We studied 10 consecutive, sedated and paralyzed patients with Acute Respiratory Distress Syndrome (ARDS). The entire study lasted 4 hours, divided in 3 periods: 2 hours of recommended ventilation [lung protective strategy, LPS, i.e., ventilation with low tidal volume (< 8 mL/kg), limiting the plateau at 35 cm H2O, together with high positive end-expiratory pressure (PEEP)], 1 hour of sigh (LPS with 3 consecutive sighs/min at 45 cm H2O plateau pressure), and 1 hour of LPS. Total minute ventilation, PEEP, FiO2 and mean airway pressure were kept constant. The introduction of sighs induced a consistent recruitment and PaO2 improvement, and a decrease in venous admixture and PaCO2. Interrupting sighs and resuming LPS led to a progressive derecruitment, and all the physiological variables returned to baseline. Derecruitment was higher in patients with higher PaCO2 and lower VA/Q ratio. We conclude that: 1) LPS alone does not provide full lung recruitment and best oxygenation in ARDS; 2) application of sigh may provide pressure enough to recruit and volume enough to prevent reabsorption atelectasis.