Daunorubucin (DNR) accumulation studies as functional tests of the multidrug resistance (MDR1) gene product P-glycoprotein have produced diverging results when correlated to response to chemotherapy in acute leukaemia. To investigate possible reasons for this diversity a starvation experiment, based upon prolongation of medium exchange, was set up in the multidrug resistant cell line CEM/VBL100. DNR accumulation (1 microgram/ml) was measured flow cytometrically in the presence or absence of Verapamil (10 micromol/l). In cells permanently kept under ideal growth conditions, addition of Verapamil resulted in an average 90% increase in DNR enhancement in five successive experiments. In contrast, DNR accumulation increased by only 26% when the medium exchange was prolonged by 30 h to 42 h. This effect was not accompanied by changes in the MDR1 gene expression at the RNA or protein level. Consequently, 53 leukaemic blast samples of 30 newly diagnosed and 18 relapsed or refractory patients with acute leukaemia (ALL-18, AML-37) were processed without any delay and under the most stringent conditions possible. Evidence of the classical MDR phenotype was arbitrarily defined by a greater than 20% enhancement in DNR accumulation in response to Verapamil (10 micromol/l) or Cyclosporin A (3 micromol/l). Using this cutoff point for analysis of newly diagnosed leukaemia we found DNR uptake better correlated to response to treatment (p = 0.002) than P-gp detection by means of immunocytochemistry, using a panel of monoclonal antibodies (p = 0.03). We conclude that DNR accumulation studies are a sensitive method for predicting therapy outcome in acute leukaemia when performed with necessary precautions.