Almost 60 randomized controlled clinical trials have been undertaken, testing the hypothesis that modulation of the endogenous host inflammatory response can improve survival for patients with a clinical diagnosis of sepsis. The results have been tantalizing, but frustrating, and no new agent has been introduced into clinical practice. Analysis of pooled data from studies of the use of an neutralizing antibody to tumor necrosis factor, or recombinant interleukin 1 receptor antagonist, show that these two approaches yield a statistically significant, but small improvement in 28 day all-cause mortality. However variability in results from one study to the next, the small absolute mortality reduction, the emerging evidence of a substantial potential for harm, and the predicted costs of recombinant biologic agents has engendered a climate of caution and pessimism. The challenge is to find methods of refining investigative approaches to maximize benefit and minimize harm. This paper reviews the recent history of sepsis clinical trials, focussing on emerging insights into the limitations of study entry criteria and measures of biologic activity and clinical benefit that may inform and direct future investigations. The biologic complexity of systemic inflammation, and the multiple interactions between clinical biology and the process of care suggest that future success in clinical research in sepsis will occur through the conduct of highly focussed investigations in a small number of dedicated centres of excellence.