In a comparative study, benzo[a]pyrene (BaP), cyclophosphamide (CP), N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) and tetrachloroethylene (PER) were tested for their ability to induce genotoxic effects in the single cell gel (SCG) test and the sister-chromatid exchange (SCE) test with human blood cells. MNNG as well as S9 mix activated BaP- and CP-induced DNA effects in both tests in a dose-dependent manner. While the range of concentrations which induced DNA migration or SCE was the same for MNNG and for BaP, much higher CP concentrations were necessary for a positive response in the SCG test than in the SCE test. PER was tested in the absence and in the presence of S9 mix and neither induced DNA migration nor increased SCE frequencies. In these experiments, a clear cytotoxic effect of PER was observed. To investigate a possible influence of DNA repair on the effects in the SCG test, cells were treated for 2 h and further incubated for 1 h after removal of the test substance. This procedure caused a clear decrease in induced DNA migration in experiments with BaP and CP, whereas no reduction was found with MNNG. This modified protocol did not lead to the detection of DNA effects after treatment with PER. The results indicate that the SCG test responds to various DNA lesions and does not seem to be sensitive to non-genotoxic cell killing. Its sensitivity obviously depends on the type(s) of induced DNA lesions and the effects can be modified by DNA repair processes in a complex manner. For the detection of genotoxic properties of chemicals with the in vitro SCG test, a single evaluation at the end of the exposure period seems to be sufficient.