To evaluate the safety and physiologic actions of ibuprofen in patients with severe sepsis. Randomized, double-blind, placebo-controlled trial. Three university hospital medical ICUs. Twenty-nine patients with clinical evidence of sepsis and the need for hemodynamic monitoring with a pulmonary artery flotation catheter. Thirteen patients received placebo and 16 received ibuprofen that consisted of 600 mg (n = 11) or 800 mg (n = 5) iv over 20 mins, followed by three 800-mg doses administered as a rectal solution every 6 hrs. The initial iv dose was given within 4 hrs of the presumptive diagnosis of sepsis. The peak circulating total ibuprofen concentration after the iv dose (49.4 +/- 4.5 micrograms/mL, mean +/- SEM) was higher than peak concentrations after the three rectal doses (17.0 +/- 2.7, 16.4 +/- 3.0, 16.0 +/- 3.1 micrograms/mL). Both routes of ibuprofen administration were well tolerated. Frequent monitoring for gastrointestinal bleeding and assessment of renal and hepatic function failed to demonstrate significant differences between ibuprofen and placebo. Because a trend for reduced creatinine clearance was observed at 8 hrs in the ibuprofen group, nephrotoxicity of this drug in sepsis cannot be excluded. Temperature decreased significantly within 4 hrs of the initial dose of investigational therapy in patients who received ibuprofen (38.5 +/- 0.3 degrees to 37.0 +/- 0.2 degrees C, p less than .001). However, despite this significant change in temperature, we were unable to detect significant differences in hemodynamic and respiratory values or survival when ibuprofen-treated patients were compared with controls. Ibuprofen was well tolerated when administered iv and rectally to patients with severe sepsis, although drug absorption was poor with the rectal route. Significant antipyretic effects of ibuprofen were demonstrated. Although an excellent safety profile characterized ibuprofen in this study, the absence of ibuprofen-associated toxicity may have been secondary to poor rectal absorption of the drug. Our results support the continued clinical investigation of ibuprofen in sepsis, using an all-intravenous route of administration.