In recent years, the testing of nanomaterials during their use in products has been increasingly emphasized, as this will provide a more realistic risk assessment (RA) compared with RA based on pristine nanomaterials. We show that (i) using such an approach for a "realistic" RA is increasing the complexity of the RA, (ii) several testing-aspects render this approach more challenging than the conventional methods, (iii) interpretation of the results becomes difficult, and (iv) the resulting RA may need to be evaluated carefully as it yields improved understanding of the short-term fate of the individual product, but long-term consequences may be neglected. (C) 2017 Elsevier Ltd. All rights reserved.