There are now unprecedented opportunities for the development of improved drugs for cancer treatment. Following on from the Human Genome Project, the Cancer Genome Project and related activities will define most of the genes in the majority of common human cancers over the next 5 years. This will provide the opportunity to develop a range of drugs targeted to the precise molecular abnormalities that drive various human cancers and opens up the possibility of personalized therapies targeted to the molecular pathology and genomics of individual patients and their malignancies. The new molecular therapies should be more effective and have less-severe side effects than cytotoxic agents. To develop the new generation of molecular cancer therapeutics as rapidly as possible, it is essential to harness the power of a range of new technologies. These include: genomic and proteomic methodologies (particularly gene expression microarrays); robotic high-throughput screening of diverse compound collections, together with in silico and fragment-based screening techniques; new structural biology methods for rational drug design (especially high-throughput X-ray crystallography and nuclear magnetic resonance); and advanced chemical technologies, including combinatorial and parallel synthesis. Two major challenges to cancer drug discovery are: (1) the ability to convert potent and selective lead compounds with activity by the desired mechanism on tumor cells in culture into agents with robust, drug-like properties, particularly in terms of pharmacokinetic and metabolic properties; and (2) the development of validated pharmacodynamic endpoints and molecular markers of drug response, ideally using noninvasive imaging technologies. The use of various new technologies will be exemplified. A major conceptual and practical issue facing the development and use of the new molecular cancer therapeutics is whether a single drug that targets one of a series of key molecular abnormalities in a particular cancer (e.g. BRAF) will be sufficient on its own to deliver clinical benefit ("house of cards" and tumor addiction models). The alternative scenario is that it will require either a combination of agents or a class of drug that has downstream effects on a range of oncogenic targets. Inhibitors of the heat-shock protein (HSP) 90 molecular chaperone are of particular interest in the latter regard, because they offer the potential of inhibiting multiple oncogenic pathways and simultaneous blockade of all six "hallmark traits" of cancer through direct interaction with a single molecular drug target. The first-in-class HSP90 inhibitor 17AAG exhibited good activity in animal models and is now showing evidence of molecular and clinical activity in ongoing clinical trials. Novel HSP90 inhibitors are also being sought. The development of HSP90 inhibitors is used to exemplify the application of new technologies in drug discovery against a novel molecular target, and in particular the need for innovative pharmacodynamic endpoints is emphasized as an essential component of hypothesis-testing clinical trials.