Overexpression of P-glycoprotein (Pgp) or MDR1 mRNA has been shown to be a negative prognostic factor for clinical outcome in acute myeloid leukemia (AML). However, resistance to chemotherapy also occurs in the absence of Pgp overexpression. Therefore, besides Pgp expression, we have assessed the expression of MRP, a novel drug transporter gene, along with the functional multidrug-resistant (MDR) phenotype of leukemic cells. These MDR parameters are correlated with clinical outcome in individual patients. We found functional changes in fresh leukemic cells from de novo or relapsed patients similar to those reported for tumor cell lines with the MDR phenotype. These changes were reduced drug accumulation as assessed with radiolabeled doxorubicin (factor 1.6), daunomycin (factor 1.13), and vincristine (factor 1.6) in patients who were refractory to the combination treatment of 1-beta-D-arabinofuranosylcytosine (ara-C) and daunomycin or mitoxantrone as opposed to patients who had complete responses. Also, the intracellular distribution of doxorubicin fluorescence (nuclear/cytoplasmic ratio), as assessed with laser scan microscopy, was reduced 1.4-fold in blasts from refractory patients. Based on historically known clinical response to single-agent daunomycin or ara-C in the group of responding de novo AML patients, we have set a threshold level such that a defined part of the samples that had the highest drug accumulation or nuclear to cytoplasmic ratios were above this threshold value. This allowed discrimination between patients responding to daunomycin from those who were refractory to this drug. By using this threshold level, in the refractory group clinical resistance corresponded with high sensitivity with a resistant phenotype. A similar threshold was set for the data of the in vitro ara-C sensitivity test. By combining both assays for all individual patients, clinical refractoriness as well as sensitivity could be predicted with high accuracy. There appeared to be no stringent relationship between the functional MDR phenotype with expression of either Pgp (fluorescence-activated cell sorting analysis) or MRP mRNA (RNase protection). However, by combining both parameters the functional MDR phenotype correlated with the overexpression of either one or both of the parameters in 94% of the samples studied. It is concluded that this combined overexpression in conjunction with functional changes for MDR drugs and ara-C reveal a correlation of MDR phenotype with clinical resistance to combination chemotherapy in AML patients and hereby may adequately predict clinical MDR in individual AML patients.