In the case of substances with a limited toxicological data base there is often (i) a lack of qualified human toxicological data; and (ii) a paucity of studies with adequate exposure duration. Hence, several extrapolations have to be performed to arrive at appropriate risk assessments or derive occupational exposure limits. The present paper deals with the possibilities for extrapolating the change in effect concentrations over time (time extrapolation, e.g. from subacute to chronic exposure) and for interspecies extrapolation (from animal to human) in connection with locally acting substances (respiratory toxicants). To justify the time extrapolation factors, 46 technical reports produced by the US National Toxicology Program (NTP) involving studies with subacute, subchronic and chronic exposure duration were evaluated. On the basis of geometric mean values, decreases in effect concentrations by factors of 3.2 (subacute --> subchronic), 2.7 (subchronic --> chronic) and 6.6 (subacute --> chronic) were found. Differentiation according to animal species (mouse, rat), sex or substance properties did not result in any relevant changes of the mean value. NTP studies with less than lifetime exposure periods (subacute, subchronic) in many cases showed different locations of respiratory effects compared with chronic studies, and thus offered limited possibilities for qualitative prediction of long-term respiratory effects (occurrence of effects in certain regions of the respiratory tract). With regard to interspecies extrapolation, gaseous and particulate substances were evaluated separately. With some modifications (e.g. consideration of the clearance of particles of low solubility), the 1994 US Environmental Protection Agency (EPA) model for deriving reference concentrations for humans on the basis of experimental data in animals is proposed for inhalable particulate substances. In the case of gaseous substances, the assumptions of the EPA model do not seem to consider sufficiently the local inhomogeneity in substance distribution and anatomical and histological differences between the upper respiratory tracts of rodents and humans. Considerable uncertainty would attach to a default factor for interspecies extrapolation for gaseous substances.