During the past few years a growing number of substances with mutagenic activity have been identified in the human environment. Such effects deserve attention not only for reasons of protecting the genetic constitution of future generations, but are also of direct concern to the present one, in view of the striking concordance between the carcinogenic and mutagenic potential of most chemicals. In the evaluation of the effects of mutagenic chemicals one can distinguish four phases: (1) primary identification or detection of mutagenic activity; (2) verification; (3) quantification; (4) extrapolation to man. For the primary identification of a particular compound as a mutagen, fast bacterial assay systems, such as those developed for Salmonella or Escherichia coli, are most suitable. Regulatory measures ought to be postponed until a verification has been obtained from at least two eukaryotic assays, such as those with Drosophila, yeast, unscheduled DNA synthesis or sister-chromatid exchanges in mammalian cells. Flexibility in the choice of tests should be maintained, since new test systems are continually being developed. Extrapolation from experimental organisms to man presents problems, since chemical mutagens are characterized by great specificity of action with regard to the spectrum of genetic changes, organisms or cell types. Thus in Drosophila, for example, most pre-carcinogens (requiring metabolic activation) produce high levels of gene mutations, but no, or only few, chromosome-breakage effects, such as translocations, dominant lethals, or chromosome loss. Even patent chromosome-breaking substances appear to require considerably higher concentrations to elicit chromosome breakage than mutations. Evidence is presented that the same situation may well be true for mammals. Since all routine mammalian assay systems (dominant lethals, cytogenetic assays, micronuclei, heritable translocations) in vivo rely on the detection of chromosome breakage, they cannot be considered as diagnostic for the induction of mutations, and may well generate false negatives. An assessment of the possible genetic hazards involved requires a quantification in terms of dose-effect curves. In the absence of suitable fast tests for gene mutations in the intact mammal, these cannot, at present, be determined. Consequently, regulatory measures have to rely on confirmatory evidence from a battery of different test systems. More extensive studies on comparative mutagenesis, to define the detection capacity, both at the qualitative and quantitative levels, for different end-points of genetic damage (gene mutation, chromosomal aberrations, non-disjunction) and malignant transformation should be given high priority. Moreover, systematic step-wise comparisons for different end-points of genetic damage, at different concentration levels, “the parallelogram”, may also help one to obtain better estimates for the induction of mutations. This approach involves extrapolations, in vitro — in vivo, that use cytogenic damage in mammalian assay systems, to calibrate for the induction of mutations, as obtained from mammalian cell lines and host-mediated assays.