Although phenol-extracted gram-negative bacterial lipopolysaccharides (LPS) have been used to study the properties of endotoxins for many years, nothing is known about the behavior of native (unextracted) LPS in vivo. Accordingly, we have compared extracted and native forms of LPS with regard to their biological activity, their ability to bind to plasma high density lipoproteins (HDL), and their fate after intravenous injection into rats. The LPS of Salmonella typhimurium G-30 were labeled with [(3)H]galactose, and whole bacteria, bacterial outer membranes, outer membrane fragments (harvested from the bacterial culture supernatant), and phenol extracts of the bacteria were prepared. After defining the LPS, phospholipid, and protein composition of these preparations, we compared the activity of the LPS in phenol extracts and membrane fragments in two assays. In both the limulus lysate assay and the rabbit pyrogen test, the LPS in phenol extracts were slightly more potent than the LPS in membrane fragments. We next studied the ability of the LPS in each preparation to bind to rat lipoproteins in vitro, and each preparation was then injected intravenously into rats for measurements of LPS-HDL binding and tissue uptake in vivo. Two patterns of lipoprotein binding were observed. Less than 25% of the LPS in both outer membranes and whole bacteria bound to HDL in vitro. When the outer membranes and whole bacteria were injected into rats, their LPS again bound poorly to HDL and they were rapidly removed from plasma into the liver and spleen. In contrast, >50% of the LPS in both culture supernatant membrane fragments and phenol-water extracts bound to HDL in vitro. When these preparations were injected into rats, approximately 50% of the LPS in the membrane fragments and phenol-water extracts bound to HDL and remained in the plasma over the 10-min study period. Moreover, the LPS in these preparations accumulated in the ovary and the adrenal gland, two tissues that use HDL-cholesterol for hormone synthesis. Binding to HDL thus greatly influenced the plasma half-life and tissue uptake of both extracted and native LPS. We conclude that extraction of S. typhimurium LPS with phenol does not significantly alter the biological activity or the lipoprotein binding behavior of the LPS and that the in vivo fates of phenol-extracted and membrane fragment LPS are essentially identical. The results thus provide important support for many previous studies that have used phenol-extracted LPS to mimic the activities of native LPS in vivo. However, the only native LPS that resembled the behavior of extracted LPS were the LPS that had been shed from the bacteria in fragments of membrane that had reduced amounts of protein and phospholipid. Removal of LPS from other outer membrane constituents, whether by chemical extraction or by a natural process of surface shedding, thus alters the behavior of the LPS; the most important feature of this alteration appears to be the ability of these LPS to bind readily to HDL.