The early stages of byssinosis, the chronic lung disorder caused by the inhalation of cotton dust, are characterized by repeated episodes of mild and reversible obstruction of airways on exposure to cotton dust. To define the relation between exposure to endotoxin and the airway response to inhaled cotton dust, we pooled and analyzed data from several previous studies of experimental exposure. The pooled data set involved a total of 108 separate sessions of exposure to dust and 32 different cottons. Each dust-exposure session had involved exposing a group of 24 to 35 prescreened healthy subjects to dust from one of the cottons for six hours. The following data were obtained for each session: average concentrations of airborne dust (range, 0.12 to 0.55 mg per cubic meter) and endotoxin (range, 6 to 779 ng per cubic meter) as determined in air samples collected by vertical elutriators, and group mean percentage change in forced expiratory volume in one second (range, +0.5 to -9.1 percent), as determined by preexposure and postexposure spirometry. When data from the 108 exposure sessions were pooled, the dust concentration was not correlated with the group mean percentage change in forced expiratory volume in one second (r = -0.08; P = 0.43). In contrast, a clear exposure-response relation was observed between endotoxin concentration and group mean percentage change in forced expiratory volume in one second (r = -0.74; P less than 0.0001). Logarithmic transformation of endotoxin values clarified this relation at low-exposure concentrations and improved the correlation (r = -0.85; P less than 0.0001). Our observations strongly support the hypothesis that endotoxin has a causative role in the acute pulmonary response to inhaled cotton dust.