The rates of loss of O6-methylguanine and O6-ethylguanine from rat liver DNA were determined over a time period of 15 min to 4 hr after various doses (5 micrograms/kg to 2 mg/kg) of dimethylnitrosamine and diethylnitrosamine which produced total amounts of these adducts in the range of 300 to 16,000 molecules/cell. This amount is considerably less than the content of O6-alkylguanine-DNA alkyltransferase protein (approximately 60,000 molecules/hepatocyte), and during the time period studied, the adducts were found to be lost with pseudo-first order kinetics. The half-life for O6-methylguanine was 47 min. O6-Ethylguanine was removed 3.6 times more slowly with a half-life of 172 min. The ability of partially purified rat liver O6-alkylguanine-DNA alkyltransferase to remove O6-methylguanine and O6-ethylguanine from [3H]alkyl-labeled DNA substrates in vitro was measured, and it was found that O6-methylguanine was removed 3.4 times more rapidly than was O6-ethylguanine. These results are consistent with the hypothesis that most, if not all, of the repair of these adducts which occurs within the first 4 hr after treatment is due to the alkyltransferase protein. Diethylnitrosamine, which is slightly more potent as a carcinogen to rat liver, produced a total amount of O6-ethylguanine of 3.7 mumol/mol guanine/mg compared to O6-methylguanine (28 mumol/mol guanine/mg) given by dimethylnitrosamine. The slower rate of loss of the ethyl adduct is not sufficient to account for this difference, and the results, therefore, support the concept that other DNA adducts (possibly O-alkylpyrimidines) contribute to the initiation of tumors by diethylnitrosamine. Preliminary evidence that the rat liver alkyltransferase can also remove hydroxyethyl groups from DNA at a rate slower than removal of ethyl groups was also obtained. Bacterial O6-alkylguanine-DNA alkyltransferase was shown to remove methyl, ethyl, and hydroxyethyl groups from the O6 position of guanine in DNA using fluorescence detection to quantitate these adducts. The bacterial protein removed methyl groups very rapidly but was much slower than the rat liver protein on the larger adducts. These results suggest that the relative rates of repair of different alkyl groups may be species specific and must be determined experimentally in the cell of interest before conclusions concerning biological effects can be drawn.