Previous analysis showed that selective inhibitors of 5 different host inflammatory mediators administered for sepsis, while beneficial with severe sepsis and high control mortality rates, were ineffective or harmful with less severe sepsis. We hypothesized that severity of sepsis would also influence inhibition of superoxide anion, another inflammatory mediator. To test this, 6 h infusions of M40401, a selective superoxide dismutase (SOD) mimetic, or placebo were given to antibiotic treated rats (n=547) starting 3 h after challenge with differing doses of intravenous E. coli designed to produce low or high control mortality rates. There was a positive and significant (p=0.0008) relationship between the efficacy of M40401 on survival rate and control mortality rates. M40401 increased or decreased the log (odds ratio of survival) (mean±sem) dependent on whether control mortality rates were greater or less than the median (66%) (+0.19±0.12 vs -0.25±0.10, p=0.01). In a subset of animals examined (n=152) at 9 h after E. coli challenge, M40401 increased (mean effect ± sem compared to control) mean arterial blood pressure (8±5 mmHg) and decreased platelets (-37±22 cellsx103/ml) with high control mortality rates but had opposing effects on each parameter (-3±3 and 28±19 respectively) with low rates (p0.05 for the differing effects of M40401 on each parameter with high vs low control mortality rates). A metaregression analysis of published preclinical sepsis studies testing SOD preparations and SOD mimetics showed that most (16 of 18) had control mortality rates greater than 66%. However across experiments from published studies, these agents were less beneficial as control mortality rate decreased (p=0.03) in a relationship not altered (p=ns) by other variables associated with septic challenge or regimen of treatment and which was similar to experiments with M40401 (p=ns). Thus in these preclinical sepsis models, possibly related to divergent effects on vascular function, inhibition of superoxide anion improved survival with more severe sepsis and high control mortality rates but was less effective or harmful with less severe sepsis. Extrapolated clinically, inhibition of superoxide anion may be most efficacious in septic patients with severe sepsis and a high risk of death.