Groups of female Wistar Furth/NCr rats, aged 6 weeks or 7 months at the start of the experiment, were administered drinking water containing N-nitrosodimethylamine (NDMA) for up to 28 days at concentrations in the range 0.2-2.54 p.p.m., resulting in daily intakes in the range 28-372 mu g/kg/day at age 10 weeks. The levels of the premutagenic DNA adduct O-6-methylguanine (O-6-meG) in liver and blood leukocyte DNA were measured at different times during this exposure as well as on the days immediately following cessation of exposure. The adduct was found to accumulate rapidly in both tissues, reaching within 2-7 days steady states in the range 0.08-0.45 mu mol/molG, similar in young and adult animals. Accumulation of O-6-meG in blood leukocytes was similar to 30% lower than in the liver. Following cessation of NDMA treatment, adducts were lost rapidly from the DNA of both tissues, with an apparent t(1/2) of similar to 19-23 h for the liver and 30-35 h for blood leukocytes, No change in liver O-6-alkylguanine-DNA alkyltransferase (AGT) took place throughout this treatment. The steady-state adduct levels were approximately linearly related to NDMA dose-rate, except that a clear break (corresponding to a 2.6-fold lower slope) was observed at dose rates >0.4 p.p.m. (similar to 56 mu g/kg/day). This dose - response relationship is in contrast to the sharp increase in the liver tumour induction in rats chronically treated with similar concentrations of NDMA reported by Pete et al. (Cancer Res., 51, 6415-6451) and suggests that accumulation of O-6-meG cannot by itself account for the hepatocarcinogenic efficacy of NDMA in the rat. Following i.g. administration of single doses of NDMA in a range approximately corresponding to the daily intake during the above mentioned chronic exposure study (25, 50 or 100 mu g/kg), repair of O-6-meG followed sharply biphasic kinetics in both liver and blood leukocytes. In the liver, an initial rapid phase (t(1/2) similar to 1.5-1.7 h) was followed by much slower repair (t(1/2) similar to 20.7-24.9 h) despite the absence of any change in AGT. Extrapolation of the two segments of the repair kinetics plots back to 0 time suggests that similar to 52-61% of the adducts originally formed in the liver and 33-35% of those formed in blood leukocytes belonged to the rapidly repaired category, The observation of biphasic repair under conditions of full AGT activity implies the existence of adduct populations with different susceptibility to repair, possibly reflecting adducts in different cell types or in different chromatin regions. The rate of loss of O-6-meG from DNA following cessation of chronic treatment was similar to that of the slowly repaired category of adducts, confirming that only adducts of this category accumulate during chronic exposure.