Bacterial endotoxin is distributed heterogeneously on the lint and plant trash in a bale of cotton. Current extraction and assay methods for lint cannot accommodate conveniently the number or amount of samples needed to adequately assess the amount of endotoxin in the bale. Estimates of endotoxin in respirable cotton dust have been poor when cotton grade, color, trash, micronaire, fiber endotoxin, and gram-negative bacteria from fiber extracts were used as individual predictors. Incomplete extraction of endotoxin from fiber and dust influences the predictability. Cold and hot water extractions were evaluated using fiber, bulk dust, and vertical elutriator (VE) dust. Hot (68 C) water extraction increased the amount of LAL assayable endotoxin as compared to cold (4 C) water extraction. Endotoxin in hot water extracts of respirable cardroom dust were compared with endotoxin in hot water extracts of separator dust from a Microtrash Monitor (MTM) and from a Shirley Analyzer. Models relating endotoxin in cardroom VE dust to the endotoxin in MTM dust in combination with color and/or trash grades, and micronaire show promise for use in assessing the endotoxic respirable dust potential of baled cotton. Hot water extracts of fiber samples may also be useful but the amounts of sample easily assayable for endotoxin are much smaller than that used for MTM dust.