Passive immunization with a human anti-endotoxin monoclonal IgM antibody (Centoxin, HA-1A) was recently studied in patients with suspected Gram-negative sepsis. Comparison of the results obtained in the Amsterdam subpopulation with those in a larger international study population of which the Amsterdam patient group was a part, showed that it had been possible to select a patient population in which HA-1A has an 'intention-to-treat' effect based upon clinical criteria (a decrease in mortality compared with placebo by 42% (p = 0.04) and in the larger study by 9% (p = 0.24). Until a clinically useful test becomes available, identification of patients who have a high likelihood of Gram-negative sepsis and who would benefit from anti-endotoxin immunotherapy with HA-1A should be based upon the history and evaluation of underlying disease, infection status, severity and progression of the disease. The severely ill patients thus selected should receive treatment as early as possible.