Chemotherapy of cancer based on cytotoxic agents has proved successful in the treatment of many cancers. The number of agents available to the oncologist has grown steadily and drug combinations are in widespread use. The perceived success of these combinations makes the introduction of new agents difficult. For any new agent, multiple phase II and III trials are likely to be needed. Since phase II/III trials usually only address single issues, the cost of introducing a new agent is substantial. Multiple studies are required with different tumor types to define the activity profile of a new drug, followed by adjusted combinations to define the role of the new drug in conjunction with older ones. Recent advances in the understanding of cancer at a molecular level are already leading to new agent design. The next problem is how to introduce and use these agents. One possible approach is to trial the drugs with tumor cells ex vivo, using a chemosensitivity assay such as the ATP-based chemosensitivity assay which is designed to mimic the situation within the tumor accurately enough to examine issues of dose response, sequence and timing in many different tumors. The avoidance of cell lines ensures relevance and the sensitivity of some of these methods allows large numbers of mechanistically logical permutations to be tested with material from small numbers of patients. The results may be used to choose the most effective combinations for clinical testing in a limited number of subsequent phase II/III trials, saving money and time, while permitting new agents to be introduced faster.