The current standard treatments for childhood cancers are highly successful. The five-year survival rates for all children diagnosed with cancer in the late 1980s approaches 70%, and the outlook continues to improve. For some types of localized embryonal tumors, such as retinoblastoma and Wilms' tumor, the cure rates approach or exceed 90%. Our success in treating childhood cancers can be attributed to the development and use of an integrated, multimodality treatment approach which includes surgery, radiaton, and combination chemotherapy. This approach has become standard treatment for most childhood solid tumors. However, for every two children who survive, one child still succumbs to cancer, and for some childhood cancers, such as neuroblastoma and certain types of brain tumors, the prognosis remains poor. Therefore, despite our successes, there remains a need to develop new chemotherapeutic agents as well as new treatment approaches for childhood cancers. In a previous volume of The Oncologist (1996;1:169-172), Dr. Charles Pratt discussed the need for developing new drugs to treat childhood cancers and the mechanisms by which the clinical trials can be efficiently conducted. However, the clinical development of new drugs and new treatment approaches becomes more difficult as our standard treatments improve. Clearly, as more patients are cured, fewer patients are available for treatment on conventional phase I and phase II trials, which are typically performed in patients who have relapsed after standard front-line and salvage therapy. The condition of patients at the time of entry onto investigational drug studies is also affected by the nature of their prior therapy. As a result of the increasing intensity of standard treatment regimens and the use of myeloablative therapy followed by bone marrow transplantation as salvage therapy, patients entering investigational drug trials are more intolerant of further treatment and they are more likely to have tumors that are refractory to any form of therapy. In essence, the children with refractory cancers who are entered on phase I and phase II trials are becoming less and less representative of children with newly diagnosed cancers. The tolerance of children and adults to anticancer drugs was reviewed by Marsoni et al. (Cancer Treat Rep 1985;69:1263-1269). Seventeen agents that entered into clinical testing prior to 1980 were studied on similar schedules in children and adults, and the pediatric maximum tolerated dose (MTD) exceeded the adult MTD for 16 of the 17 agents. For half the drugs the pediatric MTD was 30% higher than the adult MTD, suggesting that children had a significantly greater tolerance to the toxicity of cancer chemotherapy than adults. In the 1990s we have performed a number of phase I trials in children who were more heavily pretreated than those children treated on phase I trials in the 1970s, and the pediatric MTD has been equivalent to or below the adult MTD in several of our trials. Similarly, childhood cancers have been considered to be more chemosensitive than adult forms of cancer, but several recently approved agents that have significant antitumor activity in adult cancers appear to be inactive in conventional phase II trials performed in children with recurrent cancers. Careful meta-analysis of the results of phase I and II trials performed in children and adults in the 1990s will be needed to determine if the changes in the intensity of front-line and salvage treatment regimens for childhood cancers are having a significant impact on the results of conventional phase I and II trials. The clinical development of investigational drugs and the design of phase I and II may need to be adjusted to account for the apparent greater intolerance of the heavily pretreated children entering onto these trials and the greater refractoriness of their tumors. Phase I trials may need to be performed in both heavily pretreated and less heavily pretreated children, as is currently done in adults. Limited intrapatient dose escalations can also be incorporated into phase II trials, because the patient population entering phase II trials tends to be less heavily pretreated than patients entered onto phase I trials. For new agents that appear to be inactive in conventional phase II trials, additional testing can be performed in newly diagnosed patients by administering a limited number of doses of the new agent and assessing response prior to initiation of standard therapy. This phase II window has been incorporated into the design of several front-line treatment protocols. The design of future pediatric phase I and II trials must account for the changing characteristics of the patient population that will be treated on those trials. As our knowledge about the underlying molecular defects in cancer cells expands, new biologically based treatment approaches (e.g., tumor vaccines, differentiating agents, immunotherapy, growth factor inhibitors, gene therapy), which promise to provide more rational and selective therapy for childhood cancers, are being discovered. The clinical development of these exciting new approaches will also be challenging for clinical investigators. For cancers that are successfully treated with standard multimodality treatment regimens, the impact of adding one of these new treatment approaches to the standard regimen may be difficult to detect and quantify, because of the already high survival rates; and substituting a new unproven treatment approach for a highly successful standard treatment is difficult to justify. Evaluating these new treatment approaches in heavily pretreated patients with recurrent disease may also underestimate their antitumor activity. For example, tumor vaccines depend on the patient's immue response, which may be substantially suppressed by prior dose-intensive chemotherapy and radiation. It may be more feasible to initially test these new treatment approaches in those cancers in which current standard therapy is less successful. Once they have been demonstrated to have antitumor activity in poor prognosis tumors, they can be applied to the treatment of cancers that respond well to current standard multimodality treatments. With these changes in the patient population entering investigational drug studies and the development of new non-cytotoxic treatment approaches, the design and end-points of conventional phase I and II trials must be adapted to ensure that new therapies are efficiently developed in the pediatric population.