Cancer prevention can be probably obtained with easier, faster and less financial strains by pursuing educational programs aimed to induce changes in lifestyle, starting from dietary habits. In the past decades, observational and case-control studies tried to establish a functional relationship between cancer mortality and morbidity and diet. The field becomes even more intricate when scientists investigated which dietary components are responsible for the putative, protective effects of fruits and vegetables against cancer. A relevant part of the literature focused on the positive role of "antioxidant" compounds in foods, including polyphenols. The present review critically evaluate clinical and pre-clinical studies based on polyphenol administration, which contributed to support the concept, deeply rooted in the general population, that antioxidant polyphenols can fight cancer. The controversial and contradictory issues related to the pros and cons on the use of polyphenols against cancer reflect the confounding assumption that cancer treatment and cancer prevention may overlap. We conclude that a clear cut must be done between these two concepts and that the experimental approaches to investigate one or the other should be significantly different, starting from adequate and specifically selected cellular models.