OBJECTIVE - To evaluate new treatments directed against endotoxin, tumor necrosis factor alpha, and interleukin 1 for use in sepsis and related disorders (sepsis syndrome and septic shock). DATA SOURCES - Investigations of these treatments in animal models, healthy human volunteers, and patients with sepsis and related disorders. STUDY SELECTION - Particular attention was paid to studies of patients with sepsis and related disorders, especially randomized, double-blind, controlled trials. DATA EXTRACTION - Animal studies and investigations with human volunteers were judged by how closely the experimental model replicated the clinical disorder (sepsis). Patient trials were assessed by sample size and design. Results of all studies were used to evaluate the likelihood that a given treatment would reduce mortality. DATA SYNTHESIS - Direct comparison of E5 and HA-1A antibody studies is difficult because of differences in their design, definitions of shock, and methods of subgroup analysis. However, both antibodies improve outcome in some subgroups: E5 benefits patients with gram-negative infection (bacteremic or focal) who do not have refractory shock, and HA-1A benefits those with gram-negative bacteremia (regardless of whether shock is present) but not those with focal gram-negative infection. Two agents that may be beneficial in gram-positive and gram-negative infection are monoclonal antibodies to tumor necrosis factor alpha and receptor antagonists to interleukin 1. Preliminary results with both are reviewed. CONCLUSIONS - All three types of treatment may improve outcome in sepsis. The best results will probably be obtained with combination therapy that interrupts multiple points of the inflammatory cascade underlying sepsis.