Resistance to anticancer drugs is often mediated by the overexpression of a membrane pump able to extrude many xenobiotics out of the tumour cells. The most frequently expressed of these pumps is called P-glycoprotein and is encoded by a gene called MDR1 (for multidrug resistance). There could be great clinical interest for investigating the expression of this gene or of its product in patients' tumours, as well as in developing ways of circumventing this mechanism of resistance. Multidrug resistance can be diagnosed in tumours by molecular biology techniques (gene expression at the mRNA level), by immunological techniques (quantification of P-glycoprotein itself) or by functional approaches (measuring dye exclusion). Numerous studies have tried to use the MDR status of tumours as a predictor of response to treatment, but they have not yet reached definitive conclusions to allow the use of this approach in routine determinations. This is because no consensus has emerged concerning the optimal technique and the best conditions for MDR determination. Continuous efforts are still required for defining appropriate standardization of the techniques. The development of MDR modulators for the treatment of resistant tumours is a promising approach requiring rigorous clinical trials with successive phase I, phase II and phase III studies. Phase I can be omitted when the reverter is already being used in therapeutics; phase II should be performed using a sequential design, in order to prove the inefficacy of the anticancer therapy before combining it to a modulator; and phase III must only be undertaken after the demonstration that responders can be recruited by the combination. However, the effect of some reverters on anticancer drug pharmacokinetics may hamper rapid evaluation. Several drugs are good candidates for MDR modulation, but definitive results are still lacking for the introduction of such combinations in standard therapeutic protocols.