Current testing conventions for inhalation toxicity studies require that solid and non-volatile liquid compounds are converted to respirable aerosol, which is often achieved by laboratory-specific technical methodologies. So far, internationally harmonized approaches are lacking that would allow comparison of results from inhalation studies with 'contrived' test aerosols taking into account the actual particle size of the product as it might be encountered in normal handling and use. The focus of this paper is to consider aerosols of irritant substances eliciting their mode of action on sites of initial deposition within the respiratory tract of rats. Assessment is based on conventional endpoints, such as mortality (LC(50)), and sublethal endpoints that include an analysis for the concentration-effect relationship of protein in bronchoalveolar lavage fluid (BAL-protein) as a sensitive, early marker of lung edema. This retrospective analysis also addresses whether common denominators can be found for different aerosol sizes of direct and indirect irritants, such as monomeric and polymeric diphenylmethane-4,4'-diisocyanate (mMDI and pMDI), naphthylene diisocyanate (NDI), dicyclohexylmethane-4,4'-diisocyanate (HMDI), 2,4-triisopropyl-benzene-diisocyanate (TRIDI) and substances (e.g., chlorofluoroalkyl side-chain fungicides) known to decompose to irritant intermediates in the lining fluids of the airways. Collectively, this analysis shows that for irritant aerosols both the concentration and the particle size are equally important for the outcome of the test, independent of whether the endpoint chosen is lethality or BAL protein. The scientific value of 1-h exposures to high aerosol concentrations, as required by some regulations, could be challenged because high concentrations and high respirability of aerosol appear to be mutually exclusive, as shown for mMDI and NDI (LC(50 )>2000 mg/m(3)). Thus, for a meaningful risk characterization, test results from inhalation studies with 'contrived properties' due to the specific techniques employed need to be compared with the real properties of substances as marketed, handled and used.