In this paper, we discuss the derivation of respiratory health effect factors (HEFs) for use in Life Cycle Impact Assessment (LCIA) for five nanoparticles: nanosilver, nano TiO2, carbon black, high-aspect ratio, rigid MWCNT and flexible MWCNT. For this purpose, we applied the methodology of USEtox™, a consensus model to calculate human toxicity and ecotoxicity characterization factors. In view of the available data, the non-carcinogenic HEFs for carbon black and TiO2 can be attributed the recommended status. For MWCNT and nanosilver the non-carcinogenic HEFs should be awarded the indicative status due to the minimal effects observed in the animal studies. All carcinogenic HEFs are currently labelled as indicative since there is no consensus with respect to the human relevance of the overload mechanism of carcinogenicity observed in rats exposed to poorly soluble particles (carbon black, flexible MWCNT and TiO2), whereas for high-aspect ratio, rigid MWCNT, the extrapolation from asbestos carcinogenicity introduces uncertainty. For nanosilver no carcinogenic HEF could be derived. The present study clearly demonstrates the difficulties associated with the derivation of HEFs for the nanomaterials considered, in particular in the USEtox™ framework for comparative risk assessment, and makes recommendations to overcome them: • Nanoparticles for which a dose metric other than its mass is relevant for their toxicity require ‘normalisation’ to the mass metric. Adaptation of USEtox™ to accommodate other dose metrics than mass intake is therefore recommended.• Simplified estimates of deposited and retained dose are currently needed, ignoring particle size distribution dependency. Therefore, it is recommended to include kg deposited and kg retained as dose parameters for HEFs in USEtox™ and to link USEtox™ to the Multiple Path Particle Dosimetry model to convert kg intake to these dose parameters.• Establishment of the Point of Departure (PoD) for the derivation of the HEFs and the extrapolation factors to be used is recommended through consensus building. Although the ED50 is preferred as the PoD from the perspective of comparative risk assessment in LCA, for some nanoparticles, other PoDs could be considered when the dose-response data are limited.• Overall, the assessed nanoparticles are likely to be more toxic than most bulk chemicals regarding non-carcinogenic effects. However, the results of LCIA based on these HEFs must be interpreted with the necessary constraints and reservations due to the large uncertainty involved. © 2017