The concept of a “biological threshold” is attracting interest as an evaluation criterion for the mutagenic activity of DNA-targeting mutagens. In this context, the concept is defined as “a concentration of a chemical which does not produce any damage through its inability to perform the necessary biochemical reactions, even though present at the target in finite amount”. To clarify whether this criterion is indeed applicable to DNA-targeting mutagens, we re-evaluated the reverse mutation assay data using DNA repair-deficient bacterial strains, such as S. typhimurium strains lacking the O ⁶-methylguanine DNA methyltransferase genes ( ada _ST and ogt _ST), the nucleotide excision repair gene ( uvrB ) or the 8-hydroxyguanine DNA glycosylase gene ( mutM _ST), and E. coli strains lacking the nucleotide excision repair gene ( uvrA ). Mutagenic responses of 20 test chemicals including alkylating and non-alkylating agents were compared between the repair-deficient and their wild-type strains. All the alkylating agents, such as MNNG, ENNG, EMS, ENU, DMN and DEN, exhibited more sensible mutagenic responses in strains YG7108 (Δ ada _ST, Δ ogt _ST) and YG7113 (same as YG7108 but containing the plasmid pKM101) than in the parental strains TA1535 and TA100 (same as TA1535 but containing the pKM101), respectively. Upon applying MNNG, YG7108 showed about 2-100 fold increase in the number of His⁺ revertants above the spontaneous level over the range of 0.00025-0.25 μg/plate, whereas TA1535 did not show any significant increase in the number of His⁺ revertants over the same dose range. On TA1535, an increasing tendency of the number of revertants was observed at 0.5 μg/plate or above. This indicates an approximate 2,000-fold difference at the mutagenic concentration level between the wild-type and the repairdeficient strains. Other alkylating agents also showed significant differences in mutagenic responses between YG7108 and TA1535, or between YG7113 and TA100 respectively, with some variations among test chemicals. On the other hand, non-alkylating agents, such as 4-NQO, AF-2, 2-NF and MX, did not show any differences in the dose-response relationships between YG7113 and TA100. When non-alkylating agents, such as 4-NQO, 2-NF and MX were applied to TA1535 (Δ uvrB ), TA1538 (Δ uvrB ) and WP2 uvrA (Δ uvrA ), clearly different mutagenic responses, i.e. about 30- to 60-fold, were observed between the repair-deficient and the parental strains (TA1975, TA1978 and WP2, respectively). 4-NQO showed different mutagenic responses between YG3002 (Δ mutM _ST) and TA1975 (about 10-fold), though the application of other oxidative agents such as hydrogen peroxide resulted in less than 10-fold differences. The present results indicate that the wild-type strains having normal repair capacity show no gene mutation induction at the concentrations at which gene mutations are clearly induced in the repair-deficient strains through DNA damage. Thus, the present results suggest the existence of a “biological threshold” below which no mutagenic response is induced by DNA-targeting mutagenic substances.