Risk characterization comprises hazard identification describing the intrinsic toxic potential of a chemical, toxicokinetics, as well as the toxic mechanisms, information about dose response and exposure assessment. Compounds that induce reversible effects, which are repaired during and after exposure, are considered thresholded and allow definition of a NOEL. If damage is not repaired, the effect persists and accumulates upon repeated exposure. In such cases a NOEL cannot be determined. Biological reactive intermediates of chemicals have the potential to bind covalently to cellular macromolecules like proteins and DNA. Such interaction is not repaired completely and may persist. Thus, data on covalent binding (CB) are of qualitative and quantitative significance in the risk assessment process. Qualitatively, CB, especially with DNA and in correlation with this to proteins, is indicative for an irreversible and non-thresholded mutagenic and carcinogenic effect. Absence or presence of CB assists to differentiate between primarily genotoxic and thresholded non-genotoxic carcinogens. Quantitatively, CB is used to understand internal exposure and target dose, which is a prerequisite for species-species extrapolation, and to justify extrapolation from high dose to low dose. The reactive intermediates of ethylene, propylene and styrene have been determined in rodents and humans and modeled to predict dose responses of internal exposure. It is described in this communication that such information, together with other parameters like cell proliferation as a result of cytotoxicity, is the basis for quantitative risk assessment of human exposure to these compounds.