The observation that antioxidant vitamins fail to confer protective benefits in large, well-designed randomized clinical trials has led many to question the role of oxidative stress in the pathogenesis of disease. However, there is little evidence that proposed antioxidants actually scavenge reactive intermediates in vivo. Ascorbate reacts rapidly with oxidants produced by activated neutrophils in vitro, and neutrophils markedly increase their oxidant production when mice are infected intraperitoneally with the gram-negative bacterium Klebsiella pneumoniae. To explore the antioxidant properties of ascorbate in vivo, we therefore used K. pneumoniae infection as a model of oxidative stress. When mice deficient in L-gulono-gamma-lactone oxidase (Gulo(-/-)), the rate-limiting enzyme in ascorbate synthesis, were depleted of ascorbate and infected with K. pneumoniae, they were three times as likely as ascorbate-replete Gulo(-/-)mice to die from infection. Mass spectrometric analysis of peritoneal lavage fluid revealed a marked increase in the levels of oxidized amino acids and of F2-isoprostanes (sensitive and specific markers of lipid oxidation) in infected animals. Surprisingly, there were no significant differences in the levels of the oxidation products in the ascorbate-deficient and -replete Gulo(-/-)mice. Our observations suggest that ascorbate plays a previously unappreciated role in host defense mechanisms against invading pathogens but that the vitamin does not protect amino acids and lipids from oxidative damage during acute inflammation. To examine the oxidation hypothesis of disease, optimal antioxidant regimens that block oxidative reactions in animals and humans need to be identified.