Genotoxic effects observed in vitro, only at high doses or high levels of cytotoxicity, will be false positives if such conditions are not achieved or cannot be tolerated in vivo. However, for such effects to be disregarded there must be a threshold dose or level of cytotoxicity below which genotoxicity is absent. Sodium fluoride (NaF) has previously been shown to be clastogenic in vitro in Syrian hamster cells and human fibroblasts. We have extended these studies in human fibroblasts and included a positive control (mitomycin C, MMC) which is clastogenic in vivo and carcinogenic, and a chemically related control (NaCl). Cytotoxicity was measured as mitotic inhibition and cell death (loss of clonogenicity). The results are used to illustrate the problems associated with quantitative extrapolation from in vitro tests to human risk, as follows. (1) There appears to be a threshold response (clastogenicity vs. dose) with NaF at around 10 micrograms/ml (48 h exposure) but a more definitive conclusion must await elucidation of the mechanisms of clastogenicity. (2) NaCl is weakly clastogenic at 1000 times the threshold dose for NaF. The mechanisms are unlikely to be similar. (3) No clastogenicity was detected with NaF below about 30% mitotic inhibition but the relationship between clastogenicity and mitotic inhibition was similar for NaF and MMC. (4) There was no obvious threshold in the relationship between clastogenicity and cell killing with NaF. MMC was less clastogenic than NaF at equitotoxic doses. Observations 3 and 4 preclude the possibility of regarding the clastogenicity of NaF as a false positive by virtue of associated cytotoxicity.