The long-term prognostic value of tumoural MDR1 and MRP, along with p53 and other classical parameters, was analysed on 85 node-positive breast cancer patients receiving anthracycline-based adjuvant therapy. All patients underwent tumour resection plus irradiation and adjuvant chemotherapy (the majority receiving fluorouracil-epirubicin-cyclophosphamide). Median follow-up for the 54 alive patients was 7.8 years. Mean age was 53.7 years (range 28-79) and 54 patients were post-menopausal. MDR1 and MRP expression were quantified according to an original reverse transcription polymerase chain reaction multiplex assay with colourimetric enzyme-linked immunosorbent assay detection (beta2-microglobulin as control). P53 protein was analysed using an immunoluminometric assay (Sangtec). MDR1 expression varied within an 11-fold range (mean 94, median 83), MRP within a 45-fold range (mean 315, median 242) and p53 protein from the limit of detection (0.002 ng mg(-1)) up to 35.71 ng mg(-1) (mean 1.18, median 0.13 ng mg(-1)). P53 protein was significantly higher in oestrogen receptor (ER)-negative than in ER-positive tumours (P = 0.039). The higher the p53, the lower the MDR1 expression (P = 0.015, r= -0.27). P53 was not linked to progesterone receptor (PR) status, S phase fraction, or MRP Significantly greater MDR1 expression was observed in grade I tumours (P = 0.029). No relationship was observed between MDR1 and MRP. Neither MDR1 nor MRP was linked to ER or PR status. Unlike MDR1, MRP was correlated with the S phase: the greater the MRP, the lower the S phase (P = 0.006, r = -0.42). Univariate Cox analyses revealed that MDR1, MRP, p53 and S phase had no significant influence on progression-free or specific survival. A tendency suggested that the greater the p53, the shorter the progression-free survival (P = 0.076 as continuous and 0.069 as dichotomous).