The antileukemia drug amsacrine and seven analogues were tested for in vitro activity against five multidrug-resistant human leukemia cell sublines (two derived from each of two Jurkat parent lines and one from the K562 line) and the corresponding parent lines (Jurkat, K562, and P388 leukemia). Resistant cell lines were derived by exposure to either amsacrine or doxorubicin. The resistance factor was calculated as the ratio of the mean IC50 value for the resistant cell line to that for the parent line. IC50 was defined as the concentration of drug inhibiting cell growth to 50% of that in control (drug-free) cultures. Patterns of cross-resistance were visualized by plotting the deviations of resistance factors from the mean resistance factor, on a logarithmic scale. Considerable variations in resistance factors were noted for each cell subline as the amsacrine substituents were altered. Four main patterns were evident: a transport-related multidrug-resistance pattern (three sublines), a pattern similar to that for a murine P388 leukemia subline resistant to amsacrine, and two patterns not observed previously. Since some of the sublines tested showed evidence of altered topoisomerase II activity, it appears that changes in the resistance pattern in these lines may reflect changes in the stability of the ternary complexes formed by the drug, the altered enzyme, and the DNA. The resistance factor was reduced from more than 100-fold to about 13-fold in three of the sublines tested and from eightfold to twofold in two others. This finding suggests that appropriate drug design may overcome several forms of multidrug resistance in human leukemia and other types of cancer.