Nutritional epidemiology of cancer has gone through several stages. Initially, the long latency of cancer, the difficulties in undertaking long-term cohort investigations or ascertaining remote diet in case-control studies, and the absence of convenient intermediate biomarkers of disease, such as cholesterol in cardiovascular diseases, discouraged studies on diet and cancer. Subsequently, however, epidemiological successes in the chemical, viral and occupational aetiology of cancer, and the increasing insight into the sources of variation of diet and dietary information, prompted investigators to undertake both case-control and cohort studies. The results have been mixed. On the one hand, vegetables and fruits have been shown to be inversely associated with several forms of cancer. On the other hand, the information concerning specific macro- or micronutrients in relation to particular forms of cancer has been very limited and mostly inconclusive. There are several reasons for the complexity of investigations of the nutritional epidemiology of cancer and these reasons are briefly considered. An overview of our current understanding of the nutritional causes of cancer is also presented. It is noted that, notwithstanding the substantial gaps in our scientific knowledge, preventive nutritional approaches can be envisaged and they are likely to be moderately successful.