If one were to wait for the perfect set of experimental results before launching a multi-agent chemoprevention or large risk reduction study, the trial would never be launched. On the other hand, non-scientific considerations have led to the premature launching of at least three prominent studies (CARET, Carotene and Retinol Efficacy Trial; ATBC, Apha Tocopherol Beta Carotene; PCPT, Prostate Cancer Prevention Trial) and the much delayed start-up of another, BCPT, the Breast Cancer Prevention Trial. Strong epidemiologic data by itself should not be adequate to justify starting a large trial; experimental and/or clinical data should be developed. On the other hand fear of secondary adverse events that are of low incidence should not be enough to delay a trial if the overall health benefit could be high. The development of multiagent chemoprevention trials requires that each agent is active and additively or synergistically so in combination in preclinical models. Additionally, side effects of each agent should be non-overlapping and low to non-existent, preferably a feature determined in formal phase IIa and IIb trials. These principles will be discussed in the context of prior (CARET, ATBC) and ongoing (EUROSCAN, acetylcysteine/retinol), as well as proposed future trials (difluromethyl/sulindac).