Two methods, treatment in vitro and the host-mediated assay method, were compared in their ability to demonstrate the induction of MNNG of nitotic recombination in a diploid strain of Saccharomyces cerevisiae. MNNG had a strong activity in vitro but not in the host-mediated assay at the concentrations tested. When the genetic effects of MNNG have been tested in different test systems, sometimes negative, sometimes positive results have been obtained. The relevance of different tests for risk evaluation is discussed, and it is concluded from the data on MNNG that tests on whole mammals may sometimes give false negative results because the cells tested are, in parts of the body, less accessible to the mutagen. Increasing doses of MNNG by treatment in vitro gave decreasing frequencies of mitotic recombination, indicating damage to the recombinational system in the cells. Dose-response relationships for recombination and mutation are discussed.