A chromogenic Limulus amebocyte lysate assay was used to measure the recovery of 1 endotoxin unit of endotoxin per ml. Purified human high-density lipoprotein, low-density lipoprotein, and apolipoprotein A1 (apo A1) at a maximum concentration of 1 g of protein per liter reduced the recovery to less than 40% of baseline in a both dose- and time-dependent manner and in the absence of other serum components. Furthermore, the lapine fever response to a dose of 1 ml of 5-ng/ml endotoxin per kg was reduced by greater than 0.5 degrees C (P less than 0.005) when the solution was preincubated in vitro with 0.5 g of apo A1 per liter. By the Limulus test, a maximum concentration of 0.01 g of apolipoprotein B (apo B) per liter (which contained deoxycholate, a known endotoxin-disaggregating agent) reduced recovery to 0% in a dose- but not time-dependent manner. In heat-inactivated (56 degrees C, 1 h) normal human serum, high-density lipoprotein cholesterol (P less than 0.005) and apo A1 (P less than 0.05) correlated inversely with endotoxin recovery, but, paradoxically, apo B correlated directly with endotoxin recovery (P less than 0.05), while low-density lipoprotein cholesterol showed no significant correlation. INTRALIPID alone had no effect on endotoxin recovery. Addition of a maximum of 10 g of INTRALIPID per liter to 0.0042 g of apo B per liter increased endotoxin recovery from approximately 30 to 80% (P less than 0.001), but addition of INTRALIPID to 0.25 g of apo A1 per liter decreased recovery from approximately 30 to 20% (P less than 0.001). We conclude that (i) lipoproteins are endotoxin inactivators; (ii) this ability of lipoproteins may be modulated by their lipid component (lipid-endotoxin interaction); (iii) apo A1 is capable of directly inactivating endotoxin (protein-endotoxin interaction).