Résumé : Administration of sodium selenite in septic shock has been associated with apparently conflicting results that may be related to different dosing schedules. Bolus administration, leading to a transient pro-oxidative effect, could limit the inflammatory reaction and improve outcomes. We studied 21 anesthetized, mechanically ventilated, invasively monitored, and fluid-resuscitated sheep. Nine hours after inducing peritonitis by injection of autologous feces, the animals were randomized into three groups: (i) bolus injection (2 mg selenium as selenite, followed by 0.06 microg . kg-1 . h-1, n = 7); (ii) continuous infusion (4 microg . kg-1 . h-1 selenium, n = 7), or (iii) control (n = 7). No vasopressors or antibiotics were administered. All animals were monitored until spontaneous death. Peak plasma selenium values reached 4 to 14 micromol . L-1. Compared with the other groups, sheep given a bolus of sodium selenite had delayed hypotension with better maintained cardiac index, delayed hyperlactatemia, fewer sepsis-induced microvascular alterations, and a prolonged survival time (21.9 [bolus group] vs. 18.4 [continuous group] and 18.3 h [control group], P < 0.05). Hence, in this model of septic shock, the administration of a large bolus of sodium selenite (rather than a continuous administration) resulted in beneficial effects, probably by a transient oxidative effect.