It is widely accepted that tumour growth beyond a few cubic millimetres cannot occur without the induction of a new vascular supply. Inhibiting the development of new blood vessels (antiangiogenesis) is a potential approach to cancer therapy that has attracted interest in recent years. In theory, this approach should be relatively selective for tumour cells. The endothelial cells which form new vascular networks in tumours are responding to angiogenic stimuli produced by the tumour, but are themselves genetically normal. Endothelium in normal tissue, by contrast, is usually quiescent. Vascular endothelial growth factor (VEGF) is the best-characterised pro-angiogenic factor. It is virtually ubiquitous in human tumours, and higher levels have been correlated with more aggressive disease. Effective blockade of the VEGF pathway has been demonstrated with multiple agents: neutralising antibody, receptor tyrosine kinase inhibitors, and ribozyme or antisense molecules targeting expression. Promising preclinical data document the potential of these agents for tumour growth inhibition and even tumour regression, yet translation of novel therapeutics targeting the VEGF pathway to the clinic has proved a substantial challenge in itself. While showing clear evidence of antitumour activity over a broad spectrum of experimental tumours, the proper selection, dose, timing and sequence of anti-VEGF treatment in human cancer is not at all obvious. Classic Phase I dose escalation trial design may need to be modified, as higher doses may not be optimal in all patients or for all tumours. In addition, alternate or secondary biological end points (e.g., non-progression) may be needed for early phase studies to document true activity, so as not to abandon effective agents. Recent studies of the neutralising antibody bevacizumab, and small molecule tyrosine kinase inhibitor SU5416, demonstrate that, while unlikely to be effective as monotherapy, incorporation of VEGF blockade into cytotoxic regimens may increase overall response rates. However, incorporation may also produce new toxicities, including thromboembolic complications and bleeding. Newer oral agents, such as SU6668, SU11248, PTK787/ZK222584 and ZD6474, are particularly interesting for their potential for chronic therapy. Future clinical trials are likely to build on past experience with stricter entry criteria, supportive care guidelines and the use of surrogate markers.