Bacterial endotoxin interacts with the human immune system via complex immunological pathways. The evaluation of endotoxicity is important in the development of safe vaccines and immunomodulatory therapeutics. The Limulus amebocyte lysate (LAL) assay is generally accepted by the FDA for use for the quantification of lipopolysaccharide (LPS), while the rabbit pyrogen test (RPT) is used to estimate pyrogenicity during early development and production. Other in vitro assays, such as cytokine release assays with human whole blood (WB) or peripheral blood mononuclear cells (PBMCs), have also been used and may better estimate the human immunological response to products containing novel LPS molecules. In this study, WB and PBMC interleukin-6 (IL-6) and tumor necrosis factor alpha (TNF-alpha) release assays were used to estimate the endotoxic activities of purified LPS and native outer membrane vesicle (NOMV) vaccines derived from wild-type (hexa-acylated lipid A) and genetically detoxified (penta- and tetra-acylated lipid A) group B Neisseria meningitidis. A method for quantification of the differences in endotoxicity observed in the WB and PBMC assays is elucidated. The LAL assay was shown to be relatively insensitive to lipid A variations, and the RPT was less sensitive than the cytokine release assay with WB. The IL-6 and TNF-alpha assays with WB but not the assays with PBMCs distinguished between vaccines containing LPS from penta- and tetra-acylated strains. The high degree of sensitivity of the WB system to LPS variations and the presumed relevance of the use of human tissues to predict toxicity in humans suggest that this assay may be particularly well suited for the safety evaluation of vaccines and therapeutics containing acylation variants of LPS.