We tested the hypothesis that bifunctional DNA adducts formed by a nitrogen mustard-based anticancer drug were more efficient than monofunctional adducts at causing elevation of p53, consistent with the difference in cytotoxicity. Human leukemia cell line ML-1 was exposed for 1 h to melphalan or its monofunctional derivative monohydroxymelphalan. Levels of DNA adducts, measured by specific immunoassay, were linearly related to the concentration of alkylating agent. Monohydroxymelphalan formed twice as many adducts as did equal concentrations of melphalan. After the removal of the alkylating agent, adduct levels were maintained or increased slightly up to 8 h and then decreased by 27 to 44% by 24 h. Alkaline elution analyses confirmed the absence of detectable DNA interstrand cross-links in cells exposed to monohydroxymelphalan. DNA single-strand breaks were detected after monohydroxymelphalan but not after melphalan. Levels of p53 were quantified by sensitive fluorogenic enzyme-linked immunosorbent assay at intervals up to 24 h after exposure of cells to various concentrations of melphalan and monohydroxymelphalan. The level of initially formed DNA adducts needed to cause elevation of p53 from a baseline level of 0.5 ng/mg total protein to 2 ng/mg was 5-to 8-fold higher for monohydroxymelphalan than melphalan. The concentrations of melphalan and monohydroxymelphalan (ϮS.D.) causing 50% growth inhibition were 1.2 Ϯ 0.4 and 28.1 Ϯ 1.6 g/ml, respectively, a 23-fold difference. The adduct levels induced by these exposures were 9.3 and 420 nmol/g DNA for melphalan and monohydroxymelphalan, respectively, a 45-fold difference, which is considerably greater than the difference in efficacy at elevating p53.