Interest in DNA-intercalating ligands as anti-cancer drugs has developed greatly since the clinical success of doxorubicin. However, despite a great deal of 'rational design' of synthetic DNA-intercalators, only a few such compounds have proved clinically useful. This review briefly surveys the history of DNA-intercalators as clinically-used anti-cancer drugs, summarizes the known structure-experimental activity relationships and modes of action, and concludes that a factor in the slow progress is that much of the work on these compounds has been carried out by chemists, who were generally more interested in ligand/DNA interactions than drug development. Future development of the class rests on a careful consideration of the biochemical reasons behind the common limitations of the present drugs. The most important are: the inherent resistance of non-cycling cells, the rapid development (even by cycling cells) of resistance by the expression of both P-glycoprotein and altered topoisomerase II, limitations on drug distribution to and transport into tumours, low extravascular pH in tumours and the cardiotoxic side-effects of quinonoid chromophores. These considerations provide a set of constraints on physicochemical properties which must be considered in future design. However, within these constraints, there are useful future avenues for the development of DNA-intercalators as anti-cancer drugs. These include: (i) the production of improved topoisomerase inhibitors (by consideration of drug/protein as well as drug/DNA interactions); (ii) the development of reductively-activated chromophores as hypoxia-selective agents; and (iii) the use of DNA-intercalators of known DNA binding orientation as 'carriers' for the delivery of other reactive functionality specifically (sequence-, regio- and site-specifically) to DNA.