It is currently accepted that the consequences of sepsis are highly dependent on host response, largely controlled by the inflammatory reaction. Inflammation in response to infection is a continuous process from simple local reaction which regresses rapidly to systemic reaction leading to multiple organ failure. Septic shock is a specific aspect of severe infection dominated by persistent hypotension despite vascular filling. The high mortality of severe infection and shock-the incidence is about 40,000 cases per year in France -explains why new therapeutics are aimed at controlling the inflammatory reaction. The disappointing failure of several multicentric studies has placed the focus on several points. The inflammatory process is a very complex network involving a large number of subtile interactions. A given cell or a given moderator can have both beneficial and toxic effects. Results, obtained under experimental conditions quite different from human pathology should be interpreted with utmost caution before extrapolating to man. The activity of new drugs must be perfectly and completely demonstrated before starting clinical trials. The patient population in septic shock is quite varied, requiring clinical and biological markers capable of better defining those patients which could best benefit from new drugs. Much progress has been made in the methodology in large clinical trials. It is clear that a posteriori analysis of subgroups must be abandoned and that research must continue in this high-mortality pathology. The upcoming clinical trials will benefit from acquired experience.