In general there is a poor correlation between serum lipopolysaccharide (LPS; the biologically active constituent of endotoxin) levels and mortality in septic patients. The objective of this study was to determine if chemical, structural, or biological differences among LPS from different clinical isolates of gram-negative bacteria might explain this discrepancy. LPS preparations were made using the hot phenol-water extraction method from eight clinical isolates of gram-negative bacteria. As a percentage of the total weight of the LPS, the phosphate content ranged from 3.0 to 13.8% (average, 6.7 ± 3.6%), and the 2-keto-3-deoxyoctonate content ranged from 1.9 to 27.4% (average, 8.9 ± 8.5%). These values were not dissimilar to those obtained for a reference endotoxin. In a standard measure of LPS activity, the Limulus amoebocyte lysate assay, there was approximately a twofold difference between the least and most active preparations. The two preparations with the greatest difference in their ability to elicit the secretion of tumor necrosis factor alpha from a mouse peritoneal macrophage cell line were similar in lethality when administered to mice sensitized to the effects of LPS by D(+)-galactosamine. These relatively minor differences in LPS activity seem unlikely to explain the generally observed discrepancy between serum endotoxin levels and mortality in patients with gram-negative sepsis.