We prepared autologous DCs to support a Phase I clinical trial of HIV immunization in normal healthy volunteers. Final product-release testing included an assay for endotoxin, performed by the Limulus amebocyte lysate (LAL) gel clot method (Associates of Cape Cod Inc., Falmouth, MA, USA) with a cut-off of 0.06 endotoxin units/mL (EU/mL). In one recent DC preparation, the endotoxin assay was positive (0.48 EU/mL), and the product could not be released for clinical use. There were no overt signs of product contamination or bioburden; we therefore investigated other causes of the abnormal endotoxin results. We filtered all media preparations and the injectable solution for resuspension of the final product using standard laboratory sterile filters prior to use. Of note was the fact that we had used a different type of 0.2 l m filter, made of cellulose nitrate (115 mL Disposable Filterware, Nalgene Company, Rochester, NY, USA), rather than the usual synthetic polyether sulfone (PES) filter (Steriflip, Millipore, Bedford, MA, USA) for this DC preparation, and we suspected that the new filter could be the source of the contamination. We therefore repeated the LAL gel clot assays on the separate ingredients of the DC preparation — USP normal saline, autologous plasma, and the DCs themselves — before and after filtration with the Nalgene filter. Each of the ingredients was negative for endotoxin before filtration, but positive after filtration. We suspected that the ultimate source of the LAL gel clot reactivity in the cellulose nitrate filters could be