The DNA intercalating agents 4'-(9-acridinyl-amino) methanesulfon-m-anisidide (m-AMSA) and adriamycin were studied by using filter elution methods to measure DNA single-strand breaks (SSB's), DNA-protein cross-links (DPC's), and double-stranded breaks (DSB's) in mouse leukemia L1210 cells. Both compounds produced SSB's and DPC's at nearly 1:1 ratios. The SSB's and DPC's were shown to be localized with respect to each other; this was inferred from the finding that filter assays based on protein adsorption completely prevented the elution of the DNA single-strand segments between SSB's. In the case of m-AMSA, which produces relatively high frequencies of DNA lesions, the possibility that a protein bridges across the SSB was excluded by alkaline sedimentation studies. Both compounds also produced DSB's, but the SSB/DSB ratios differed; the SSB/DSB ratios increase in the following order: ellipticine greater than adriamycin greater than m-AMSA greater than X-ray [results of this paper combined with those of Ross, W. E., & Bradley, M. O. (1981) Biochim. Biophys. Acta (in press)]. The o-AMSA isomer is much less cytotoxic than m-AMSA and did not produce protein-associated strand breaks. The simplest model to explain the results is that a protein becomes covalently bound to either the 3' or the 5' termini of the intercalator-induced strand breaks. At moderately cytotoxic doses, m-AMSA yielded much larger frequencies of protein-associated SSB's than did adriamycin. m-AMSA-induced protein-associated SSB's saturated at approximately 60000 per cell over a concentration range in which m-AMSA uptake by the cells was proportional to the drug concentration. m-AMSA was found to enter and exit from cells very rapidly at 37 degrees C; protein-associated SSB's and DSB's also appeared and disappeared rapidly. At reduced temperature, however, the appearance and disappearance of protein-associated SSB's could be blocked while m-AMSA entry and exit still occurred. The saturation behavior and temperature dependence suggest that the formation and disappearance of protein-associated strand breaks is enzymatic. The simplest hypothesis is that the linked protein is a nuclease, such as a topoisomerase, which becomes bound to one terminus of the strand break it produces. It is proposed that topoisomerases producing SSB's and DSB's are stimulated to different degrees by different intercalators.