Issues of biological relevance and thresholds for genotoxicity are discussed here based upon the background of experience with the submissions for the approval of new pharmaceuticals to the German regulatory authority over the period between 1990 and 1997. This experience shows that out of the genotoxicity test systems which are required according to existing guidelines in the European Union (EU), the in vitro tests for chromosomal aberrations (CA) and the mouse lymphoma tk assays (MLA) yield a rate of positives that is about four-fold higher than that of other genotoxicity tests. A detailed analysis of chemical and pharmacological classes of compounds and their effects in these systems reveals that in addition to direct DNA reactivity several mechanisms of indirect genotoxicity such as nucleoside analogue incorporation into DNA, interaction with microtubule assembly, topoisomerase inhibition and high levels of cytotoxicity are relevant. New pharmaceuticals, for which the latter mechanisms apply, often display threshold-like characteristics in their genotoxic effects in vitro or even in vivo in experimental animals. This casts doubt upon the relevance of positive in vitro test results for such compounds. However, the discussion of examples shows that it may not be easy to demonstrate the exact thresholded mechanism of genotoxicity in a given case. In particular, the demonstration of a coincidence of genotoxicity and high levels of cytotoxicity, which seems to be a major factor for biologically non-relevant in vitro positive new pharmaceuticals, usually requires quite extensive testing. Hence, for new pharmaceuticals it is practice to provide in addition to in vitro results that may be thresholded a wealth of information from in vivo studies on genotoxicity, carcinogenicity, metabolism, pharmacokinetics, etc. the results of which help in assessing the biological relevance of in vitro positives. The regulatory acknowledgement of biologically non-relevant, thresholded mechanisms of (in vitro) genotoxicity in addition to those that are considered relevant for human risk ensures a better understanding of test results and is needed for the credibility of genotoxicity testing practice in general.