The genotoxic effects of three chlorohydroxyfuranones (CHFs), 3-chloro-4-(dichloromethyl)-5-hydroxy-2[5H]-furanone (MX), 3-chloro-4-(chloromethyl)-5-hydroxy-2[5H]furanone (CMCF) and 3,4,-dichloro-5-hydroxy-2[5H]furanone (MCA), which are formed as byproducts of water disinfection with chlorine, were investigated in bacterial differential DNA repair assays in vitro and in animal-mediated assays in vivo. As indicators of DNA damage, E. coli K-12 strains were used that differ in their repair capacity (uvrB/recA vs. uvr+/rec+). Liquid incubation of the compounds without metabolic activation caused a pronounced reduction of the viability of the repair-deficient strain relative to the repair-proficient wild-type strain. The order of potency of genotoxic activity in vitro (dose range 0.004-10 micrograms/ml) was MX > CMCF > MCA. Addition of mouse S-9 mix or bovine serum albumin to the incubation mixtures resulted in an almost complete loss of the activity of all three test compounds. In the animal-mediated assays, mixtures of the indicator bacteria were injected intravenously into mice which were subsequently treated with the test compounds (200 mg/kg b.w.). Two hours later, the cells were recovered from various organs and the relative survival frequencies determined. Under these conditions, all three compounds caused pronounced genotoxic effects, MX and CMCF being stronger genotoxins than MCA. The strongest effects were consistently found in the gastrointestinal tract, but statistically significant DNA damage was also observed in indicator cells recovered from lungs, liver, spleen and kidneys. In a further experiment, the effects of lower doses of MX (4.3, 13 and 40 mg/kg) were investigated. In these experiments dose-dependent effects were measured in all organs. CMCF and MA caused only marginal effects at 40 mg/kg except in the stomach where approximately a 50% reduction of relative survival frequency was observed with CMCF. The results of these animal-mediated assays indicate that (i) all three CHFs cause genotoxic effects in the living animal, and (ii) the potencies of the three compounds observed under in vivo conditions are not commensurate with their extremely high activities measured in vitro. One possible explanation for the weaker responses observed in the animal-mediated assays might be that CHFs are inactivated by non-specific protein binding.