The two formaldehyde (FA)-releasers dimethylol urea (DMU) and diazolidinyl urea (DZU) are widely used as preservatives or additives. They were tested for genotoxicity in three short-term test systems, i.e. in the Salmonella typhimurium mutagenicity assay, in the in vitro micronucleus test with V79 Chinese hamster cells and in the in vitro tubulin assembly assay using isolated tubulin from pig brains. The polymerization products obtained in the tubulin assembly assay were examined additionally by electron microscopy. In the S. typhimurium mutagenicity assay with the pre-incubation assay both FA-releasers tested show a clear and concentration-dependent increase in the number of revertants in strains TA98, TA100 and TA102 with and without metabolic activation (rat liver S9 mix). In all cases, a biologically relevant increase in the number of revertants was achieved within the concentration range tested (DZU: 0.04-1.8 micromol per plate, DMU: 0.21-8.33 micromol per plate). FA was tested at 0.06-2.5 micromol per plate and lead to similar effects.Both compounds induce the formation of micronuclei (concentration range tested: DZU: 2.5-50 micromol/l, DMU: 3.3-333 micromol/l). However, DMU shows a comparatively weaker effect exclusively in the absence of the metabolizing enzymes. By contrast, DZU yields a distinct increase of the micronucleus rate in the absence and in the presence of S9. In addition, DZU predominantly causes an increase of large micronuclei, which suggests that this compound has a marked aneugenic potential. Cytotoxic effects accompany the clastogenic effects of both DMU and DZU. The examination of DMU and DZU in view of a possible aneugenic potential in the tubulin assembly assay yielded the following results: DMU at concentrations up to 10 mmol/l did not influence the formation of microtubuli, whereas DZU inhibited this process completely at 3 mmol/l. FA at 6 mmol/l completely inhibited the tubulin assembly. These results could clearly be confirmed by electron microscopy examination. The different potential of the two compounds with respect to the inhibition of tubulin formation is apparently due to a significant difference in the degree of FA release. According to these results, both compounds have to be considered as genotoxic in vitro. On account of these data and because of the widespread use of these two compounds in various products used in daily life, a reevaluation of the risk associated with these compounds seems to be necessary.