Epidemiological studies have identified an inverse relationship between ingestion of plant foods and cancer risk. However, only ∽2/3 of such studies show this association. Clinical trials based on epidemiological findings require preclinical studies to provide insight into reproducibility. The beta carotene story is an example of clinical trials based on epidemiological data, before mechanism, dose or the bioactive component had been clearly identified. Results showed rather than prevention, an increase in lung cancer in smokers. Epidemiological studies are used successfully to generate hypotheses for in vitro mechanistic studies of isolated components from plant foods, such as sulforaphane from broccoli. Yet even these studies are insufficient to plan clinical trials of whole foods, since bioavailability, disposition, dose, and effects of the food matrix remain unknown. Evidence-based information, from animal and small clinical studies carried out prior to clinical trials can assure an optimal design. Research into effects of broccoli and sulforaphane make an excellent example of how data gaps have closed between epidemiology and clinical trials. Data on efficacy of broccoli in animal cancer prevention studies are strong, and small clinical studies are emerging. The time is right for clinical trials of purified and semipurified sulforaphane, as well as whole broccoli.