Almost 30 years after the unsuccessful clinical evaluation of camptothecin sodium, there has been a revival in interest in this class of agent that poisons topoisomerase I. Currently there are four camptothecin analogues in clinical trials each at different levels of advancement. Clinical data suggest that patterns of antitumor activity and toxicity profiles differ between analogues. In preclinical models antitumor activity appears to be highly schedule-dependent. Here we review rodent and human tumor models used in evaluation of efficacy, and models used to predict toxicities of these compounds. The major limitation of rodent models is that the mouse tolerates significantly greater systemic exposure to each camptothecin analogue than do patients. This leads to a false overprediction of potential clinical activity. However, responses of human tumor xenografts in mice are highly predictive of responses of clinical cancer when camptothecins are administered at dose levels achieving similar systemic exposure in mice. Development of assays that identify analogues that maintain therapeutic activity in mice, but have less species differential toxicity, particularly to the hematopoietic system, may provide an early screen to select compounds having greater clinical utility.