Dose-response studies were performed with the alkylating agents [nitrogen mustard, N,N'-bis(2-chloroethyl)-N-nitrosourea, melphalan, cisplatin (CDDP), 4-hydroperoxycyclophosphamide (4-HC), and trimethyleneiminethiophosphoramide] in both the MCF-7 human breast carcinoma cell line and the EMT6 and FSaIIC murine tumor lines. Increasing selection pressure with the alkylating agents CDDP, melphalan, and 4-HC in vitro produced low levels (6.5- to 9-fold) of drug resistance, despite an intensive and prolonged treatment program. The MCF-7 sublines made resistant to CDDP and 4-HC did not exhibit cross-resistance to other alkylating agents; however, the MCF-7 subline resistant to melphalan was partially cross-resistant to nitrogen mustard, 4-HC, and CDDP. A log-linear relationship was maintained between surviving fraction of MCF-7 cells in culture and drug concentration with alkylating agents, whereas for nonalkylating agents the survival curves tended to plateau at high drug concentrations. Log-linear tumor cell kill was also obtained over a wide dosage range with several alkylating agents in murine tumors treated in vivo. Tumor cell survival assay by colony formation indicated the continuing importance of dose in the action of the drugs even at high levels of tumor cell kill. With some agents, there was a difference between the slopes of the tumor cell killing curves in vivo as compared to in vitro. Cyclophosphamide was far more potent in vitro (4-HC) than in vivo (cyclophosphamide). Trimethyleneiminethiophosphoramide and N,N'-bis(2-chloroethyl)-N-nitrosourea were both more potent in vivo than in vitro. These differences may be explained by the various metabolic patterns of these drugs. Dose of alkylating agents is clearly a crucial variable particularly where multilog tumor cell kill is the goal, and in this regard, the effect of drug dose on the tumoricidal action of the alkylating agents is substantially greater than for nonalkylating agents.