Interleukin 8 (IL-8), a potent activator of neutrophils, may be important in the early host response to serious Gram-negative infections. IL-8 was measured with other acute phase cytokines (tumor necrosis factor alpha [TNF-alpha], IL-6 and IL-1 beta) in 25 normal humans randomized to receive either intravenous endotoxin alone or endotoxin after oral administration of ibuprofen or pentoxifylline, agents that alter some of the inflammatory responses induced by endotoxin in vitro. TNF immunoreactivity was maximum at 1.5 h, and total TNF (area under the curve) was 4.2- and 4.5-fold greater in subjects given endotoxin/ibuprofen compared to subjects given endotoxin alone (p = 0.026) or endotoxin/pentoxifylline (p = 0.004), respectively. IL-6 levels were maximum at 2-3 h and did not differ among the three groups. No IL-1 beta was detected in any subject. IL-8 levels peaked at 2 h in subjects given either endotoxin alone or endotoxin/pentoxifylline, falling towards baseline by 5 h. Subjects given endotoxin/ibuprofen had a more sustained rise in IL-8 with peak levels 2.8- and 2.5-fold higher at 3 h compared to endotoxin alone (p = 0.048) or endotoxin/pentoxifylline (p = 0.023), respectively. Differences in total IL-8 release among groups approached statistical significance (ANOVA, p = 0.07). This trend reflected the increased release of IL-8 by the subjects receiving ibuprofen compared to pentoxifylline (1.9-fold higher; p = 0.024). This suggests that cyclooxygenase products may provide important negative feedback loops for cytokine production in vivo. Increases in circulating IL-8 are part of the acute inflammatory response of humans to endotoxin. Altered cytokine responses caused by antiinflammatory therapy may have important implications for both host defense and injury during septicemia.