Abstract
Recent disappointing results have challenged the earlier results of ecological studies that launched the hypothesis of a strong relationship between diet and cancer. The present state of knowledge regarding the relationship between cancer and diet is considered and discussed here. Steps for improving the understanding of the relationship and the content of recommendations for cancer prevention and survival are proposed, such as determining the possible food effect at each step of the carcinogenesis process, considering the dietary pattern instead of a single nutrient or food, introducing the diet quality index for evaluating cancer risk and developing more comprehensive statistical methods in nutritional epidemiology. In support of these propositions, previous, recent and on-going studies are reviewed and discussed. A holistic model of diet is described as a conclusion.