Mutagenic substances classified as carcinogens are primarily regulated on the basis of their carcinogenic effect. Regulation of mutagens that have not been tested for carcinogenicity represents a problem. In cases where a threshold cannot be identified, the substances may be banned or if their uses are deemed to be unavoidable, the exposure may be reduced to as low as technically and economically feasible. In an attempt to develop a procedure that may be helpful in regulation of mutagenic substances when studies on carcinogenicity are lacking, we have compared the lowest effective dose (LED) giving a response in an in vivo genotoxic test after oral or inhalation exposure with the carcinogenic dose descriptor T25 (the chronic daily dose which will give 25% of the animals tumours above background at a specific tissue site). The 34 carcinogens in the present analysis for which genotoxic mechanisms are likely or possible, represent different classes of carcinogens and different genotoxic endpoints, exhibiting carcinogenic and mutagenic potencies both covering a range of 10,000 between the most and least potent substances. A linear correlation was found between the lowest effective dose for in vivo genotoxicity after oral administration or inhalation exposure and the lowest dose descriptor T25 for tumour formation. The finding that the median of the ratio LED/T25 was 1.05 and that the ratio for 90% of the substances studied fell in the range 0.21 to 9.2 shows that the numerical value of LED is similar to the numerical value of T25 within a factor of 5-10. The results suggest that LED may be used as a basis for regulation of mutagens in cases where a threshold cannot be demonstrated or inferred, and where the substance has not been studied in long-term carcinogenicity studies. In such cases LED divided by a specified assessment factor may represent a virtually safe level or a tolerable risk level for a possible carcinogenic effect.Genotoxic effects are assumed to play an important role in tumour induction of a number of chemicals. Since possible carcinogenic effects have been considered to be more critical than germ cell mutagenesis with regard to chemical mutagens, mutagenic substances classified as carcinogens are primarily regulated on the basis of their carcinogenic effect. Several methods for quantitative hazard and risk characterisation (e.g. the linearised multistage (LMS) model (US EPA 1986), the LED10 method (US EPA 1996), and the T25 method (Sanner et al. 2001) have been used in the regulation of carcinogens.Regulation of mutagens that have not been tested for carcinogenicity, represents a problem. In cases where the genotoxic effect is presumed to exhibit a dose threshold, and the description of the threshold is considered to be robust both from a mode of action point of view and from the available experimental results, a margin of safety approach may be recognised as appropriate for regulation. In cases where a threshold cannot be identified, the substances may be banned or if their uses are deem...