Endotoxin in house dust has been shown to be associated with asthma severity. Little is known about the influence of housing characteristics on endotoxin distribution. Using standardized methods, dust was sampled from a 1m(2) site and the whole accessible carpet area in selected Wellington, New Zealand homes (n = 77). Endotoxin was measured using a Limulus Amoebocyte Lysate assay. Relative humidity and temperature were recorded using sensors placed in carpet bases. Questionnaires were used to collect information on housing characteristics. All analyses were performed for endotoxin units (EU)/mg and EU/m2 for each site. Geometric mean endotoxin levels were 22.7 EU/mg [geometric standard deviation (GSD) = 2.4] or 30,544 EU/m2 (GSD = 3.2) from the 1m(2) site, and 28.4 EU/mg (GSD = 3.4) or 5653 EU/m2 (GSD = 6.4) from the whole room. After controlling for confounding, endotoxin was positively associated with dogs inside [geometric mean ratio (GMR): 0.9-2.0], total household occupants (GMR: 1.7-2.0, for 1 m2 sample only), vacuum cleaners <1-year old (GMR: 2.3-2.7), reusing vacuum dust collection bags (GMR: 1.4-3.1), steamcleaning or shampooing the carpet (GMR: 1.4-2.2) and high relative humidity (GMR: 1.4-1.6). Lower endotoxin was associated with floor insulation (GMR: 0.4-0.8), and north-facing living rooms (GMR: 0.4-0.8). This study has identified home characteristics that could be modified to reduce endotoxin exposure.