AIMS--To determine the optimal working conditions of the alkaline phosphatase-antialkaline phosphatase (APAAP) method to establish a specific and sensitive assay for the detection of low numbers of MDR positive cells in patients with hematological malignancies. METHODS--Three monoclonal antibodies (C-219, JSB-1, MRK-16) were used for the detection of P-glycoprotein (P-gp) in cell lines and in samples from 43 patients with haematological malignancies. The results of the APAAP method were compared with western blotting for specificity and sensitivity. RESULTS--Excellent correlation was obtained between optimised APAAP and western blotting, except in the case of multiple myeloma. JSB-1 seemed to be the more useful monoclonal antibody for the APAAP which was more sensitive than western blotting in its ability to detect single P-gp positive cells. CONCLUSIONS--Methods for P-gp detection, as defined by multidrug resistant (MDR) cell lines, are not necessarily optimal and specific for clinical samples and may lead to higher false positive and negative results, according to the conditions and the monoclonal antibodies used.