Cancer chemotherapy remains the only treatment modality with curative activity against multiple forms of metastatic malignancy. Over the past decade, cytotoxic and anti-endocrine drugs have been supplemented by targeted therapies that seek to exploit the molecular lesions that underlie the carcinogenic process or maintain the cancer phenotype. Success with, for example, Imatinib and Trastuzumab has suggested that identification and validation of the drug target is the starting point for the optimal route to the development of active drugs. However, in reality, our understanding of the biology of cancer is still too rudimentary to allow drug developers to rely on the simplistic linear pathway of target identification and validation, lead identification and optimisation, followed by Phase I, II and III trials. As pre-clinical and clinical drug developers investigate the second wave of targeted agents, it is worthwhile reflecting on experience gained during the initial development of cytotoxic drugs. For example, the clinical activity of alkylating agents and antimetabolites was demonstrated before the targets for these drugs were defined in any detail. Recent experience with signal transduction modifiers has again shown that agents initially developed to exploit one target may actually hit other targets, and that interaction with these other targets may be responsible for the clinical activity of the compound. Using lung cancer, the world's single biggest cancer problem, as an example the development of recently evaluated drugs, both cytotoxic and targeted, is reviewed. On the basis of this Review, it is concluded that drug developers should design pre-clinical studies and early clinical trials in a manner that allows both the pharmacology of the drug as well as the biology of the target to inform the development process.