Conventional models for successful chemotherapy suggest that early-stage ovarian cancer should be more responsive to cytotoxic drugs than advanced disease, that multiple drugs will prove more efficacious than single agents and that more intensive chemotherapy will prove superior to conventional doses. The purpose of numerous clinical studies has been to test these predictions and the results often fall short of the expectations. In trials to date, adjuvant chemotherapy for high-risk, early-stage ovarian cancer has provided only a modest, equivocal increase in disease-free survival. Combination chemotherapy produces higher response rates but this has not consistently resulted in prolonged survival. Dose intensification with high-dose chemotherapy increases rate of response further, but most responses are of short duration. The objective of this review is to integrate clinical and molecular biological observations into a novel model of drug resistance. We hypothesize that drug sensitivity is an acquired characteristic of neoplastic cells, and that there are two broad cellular forms of resistance to cytotoxic drugs. 'Physiological drug resistance' is a cellular state when drug sensitivity of cancer cells is similar to that of the corresponding normal tissue. This form of drug resistance can precede the acquisition of drug sensitivity and may be predominant in the early stages of neoplastic progression. A distinct, 'pathological drug resistance' can reappear later during tumor progression or during chemotherapy as a result of increased detoxification, upregulation of repair pathways or defective apoptosis. Tumors are formed of heterogeneous cell populations and in terms of drug sensitivity three distinct cell types may be present: drug-sensitive, physiologically drug-resistant and pathologically drug-resistant. Clinical response to chemotherapy is determined by the relative contribution of these different cell populations to the total cellular mass. Depending on the predominant type of surviving population, distinct patterns of clinical failure may develop that require different treatment strategies for optimal management. For chemosensitive cells that survived insufficient chemotherapy, consolidation with further, perhaps, high-dose chemotherapy is a rational option. For pathologically drug-resistant cells, pharmacological manipulation of drug resistance, and signaling pathways in combination with chemotherapy could be exploited. For physiologically drug-resistant cells, non-chemotherapy-based, 'chemopreventive' strategies to arrest tumor progression may prove beneficial.