OBJECTIVE: We investigated whether hypoxemic resuscitation from hemorrhagic shock prevents the late circulatory instability and attenuates the oxidative and inflammatory responses associated with the standard strategy. DESIGN AND SETTING: Prospective, randomized, controlled animal study in an experimental laboratory of a university intensive care unit. SUBJECTS: Thirty-one New Zealand white rabbits weighting 3.1-3.4 kg INTERVENTIONS: Anesthetized animals were subjected to hemorrhagic shock by exsanguinations to a mean arterial pressure of 40 mmHg for 60 min. Resuscitation was performed by reinfusing the shed blood for 30 min under normoxemia (PaO(2) 95-105 mmHg, control group, n=10) or hypoxemia (PaO(2) 35-40 mmHg, hypox-res group, n=10); Ringer's lactate was given from 30 to 60 min to restore arterial pressure within baseline values. A sham group was assigned (n=11). Animals were recorded for 120 min postresuscitation and for further 360 min to assess the early mortality rate. MEASUREMENTS AND RESULTS: Hypoxemic resuscitation compared with normoxemic resuscitation from hemorrhagic shock was associated with (a) a better hemodynamic condition assessed by the gradual restoration of blood pressure, higher urinary output associated with less fluid infusion; (b) lower reactive oxygen species production assessed by the reduced blood geometric mean fluorescence intensity, lower malondialdehyde, and higher ratio of reduced to total glutathione levels; (c) attenuation in the plasma concentrations of IL-1beta, TNF-alpha, and IL-6; and (d) no difference in mortality rate. CONCLUSIONS: Hypoxemic resuscitation from hemorrhagic shock is more efficient than normoxemic in restoring the blood pressure and in attenuating the excessive oxidative and inflammatory responses observed during normoxemic resuscitation.