1. Methods were developed for analysis of alkylpurines, O2-alkylcytosines, and representative phosphotriesters [alkyl derivatives of thymidylyl(3'-5')thymidine], in DNA alkylated in vivo, using high-pressure liquid chromatography. 2. The patterns of alkylation products in DNA in vivo at short times were closely similar to those found for reactions in vitro. Alkylation by the nitrosoureas was complete in vivo within 1 h, but with ethyl methanesulphonate was maximal at 2--4h. 3. The time course of persistence of alkylation products in vivo was determined for several tissues. In addition to the rapid loss of 3- and 7-alkyladenines reported previously for all tissues, a relatively rapid loss of O6-alkylguanines from DNA of liver was found which was more rapid at lower doses. In brain, lung and kidney, excision of O6-alkylguanine was much less marked, but was not entirely excluded by the data. In thymus, bone marrow and small bowel, all alkylated bases were lost with half-lives of 12--24h, at non-cytotoxic doses of alkylation. 4. No evidence for any marked excision of other minor products from alkylated DNA in vivo was found; thus 1-methyladenine, O2-ethylcytosine (found in appreciable amount only with N-ethyl-N-nitrosourea), 3-methylguanine, and dTp(Alk)dT persisted in alkylated DNA, including DNA of liver. 5. The induction of thymic lymphoma was determined over the range of single doses by intraperitoneal injection up to about 60% of the LD50 values, and related to the extent of alkylation of target tissues thymus and bone marrow. With N-methyl-N-nitrosourea over 90% tumour yield was attained at 60 mg/kg, and with N-ethyl-N-nitrosourea up to 52% at 240 mg/kg, but with ethyl methanesulphonate at up to 400 mg/kg only a few per cent of tumours were obtained. 6. The carcinogenic effectiveness of the agents was positively correlated with the extents of alkylation of guanine in DNA of target tissues at the O-6 atom. On the basis that at doses giving equal carcinogenic response these extents of alkylation would be equal, the chemical analyses showed that the ratio of equipotent doses to that for N-methyl-N-nitrosourea would be, for N-ethyl-N-nitrosourea, 5.3 for ethyl methanesulphonate about 21, and for methyl methanesulphonate [Frei & Lawley (1976) Chem.-Biol. Interact. 13, 215--222] about 144. These predictions were in reasonably good agreement with the observed dose-response data for these agents.