Abundant preclinical and clinical data suggest that cancer is a preventable disease; however, few, if any, of the medical interventions available are completely without adverse effects or risk to the patient. This statement is especially true with regard to phenotypically healthy but high-risk populations. Balancing the risks from interventions with the risk of future life-threatening disease is particularly challenging because no standardized methodologies to perform such calculations exist. It is critical, therefore, to establish a framework for determining when interventions show sufficient promise of efficacy and sufficient safety to justify their testing in clinical trials. Systematic review of all available preclinical, epidemiological, and clinical data, along with a mechanistic understanding of the biology of the disease under study, is mandatory before clinical trials are embarked upon. This Review identifies the issues that are critical for decision-making when clinical trials in human beings are being contemplated and provides a framework that can be applied in making these decisions.