Exposure to farming environment in early life has been associated with lower risk for allergic diseases possibly caused by increased exposure to endotoxin. The aims of this study were to compare the reproducibility of different sampling methods for endotoxin, and to determine whether environmental characteristics have different effect on endotoxin levels of different sample types. The reproducibility of sampling methods (bed dust, floor dust, vacuum cleaner dust bag dust, settled dust and air samples) was studied with repeated sampling (five visits during 1 year) in five farming and five urban homes. To examine determinants of endotoxin for different types of dust sample, sampling was conducted once in 12 farming and 17 urban homes. Endotoxin was analyzed using Limulus Amebocyte Lysate assay. Bed dust samples had the best reproducibility (intraclass correlation, ICC=66%), but the difference between farming and non-farming homes was not clear with this sample type. The reproducibility of floor (ICC=52%) and settled dust (ICC=51%) was moderate. With these sample types the difference between farming and non-farming homes was clear. Settled dust had some seasonal variation. Based on this study, the best compromise for sampling for endotoxin appears to be floor dust sample followed by bed and settled dust samples. Practical Implications Endotoxins have been widely measured, even though the validity of different sample types to reflect the endotoxin exposure level of an indoor environment is poorly known. This study shows that bed dust samples have the best reproducibility, but they do not reflect the differences in exposure due to environmental factors such as farming. Floor dust samples with moderate reproducibility may be the best choice for sampling of endotoxin in large field studies.