In spite of our expanding knowledge on the molecular biology of cancer, relatively little progress has been made in improving therapy for the solid tumours which are major killers, e.g., lung, colon, breast. Significant advances over the past 10-15 years in chemotherapy of some tumours such as testicular cancer and some leukaemias indicates that, in spite of the undesirable side-effects, chemotherapy has the potential to effect cure in the majority of patients with certain types of cancer. Multidrug resistance, inherent or acquired, is one important limiting factor in extending this success to most solid tumours. In vitro studies described in this review are now uncovering a diversity of possible mechanisms of cross-resistance to different types of drug. Sensitive methods such as immunocytochemistry, RT-PCR or in situ RNA hybridisation may be necessary to identify corresponding changes in clinical material. Only by classifying individual tumours according to their specific resistance mechanisms will it be possible to define the multidrug resistance problem properly. Such rigorous definition is a prerequisite to design (and choice on an individual basis) of specific therapies suited to individual patients. Since a much larger proportion of cancer biopsies should be susceptible to accurate analysis by the immunochemical and molecular biological techniques described above than to direct assessment of drug response, it seems reasonable to hope that this approach will succeed in improving results for cancer chemotherapy of solid tumours where other approaches such as individualised in vitro chemosensitivity testing have essentially failed. Results from clinical trials using cyclosporin A or verapamil are encouraging, but these agents are far from ideal, and reverse resistance in only a subset of resistant tumours. Proper definition of the other mechanisms of MDR, and how to antagonize them, is an urgent research priority.